=Paper=
{{Paper
|id=Vol-2996/paper2
|storemode=property
|title=Field Lab: An Intelligent Ecosystem for Longitudinal Design Research
|pdfUrl=https://ceur-ws.org/Vol-2996/paper2.pdf
|volume=Vol-2996
|authors=Peter Lovei,Eva Deckers,Mathias Funk,Stephan Wensveen
|dblpUrl=https://dblp.org/rec/conf/ewsn/LoveiDFW21
}}
==Field Lab: An Intelligent Ecosystem for Longitudinal Design Research==
https://ceur-ws.org/Vol-2996/paper2.pdf
Field Lab Sleep and Energy: A System for Longitudinal Remote
Sleep Tracking and Prototyping
Peter Lovei Mathias Funk
Eva Deckers Stephan Wensveen
peter.lovei@philips.com m.funk@tue.nl
eva.deckers@philips.com s.a.g.wensveen@tue.nl
Philips Experience Design Industrial Design, Eindhoven University of Technology
Eindhoven, The Netherlands Eindhoven, The Netherlands
Figure 1: The Field Lab Intelligent ecosystem illustrated by a participant looking at the Field Lab Sleep mobile application
(left), and two data visuals that are based on data collected by the shown Wakeup Light [9] and the Withings Sleep Analyzer
[15] (right)
ABSTRACT CCS CONCEPTS
There is both clinical and consumer interest in remote sleep track- • Human-centered computing → Systems and tools for inter-
ing technologies. Sleep is interesting to be observed remotely, in action design; User studies.
a (smart) home environment. For longitudinal design research we
needed to collect data for a longer period of time. Therefore, us- KEYWORDS
ing the Data-Enabled Design process we built Field Lab Sleep and
data-enabled design, longitudinal design research, human-IoT ex-
Energy. We conducted a user study for a period of one year using
periences, prototyping, designing for sleep
this system. Using the system the team could gather, store, process,
visualize and analyze the incoming behavioral, experiential and
contextual data. We built and embedded a communication plat-
form in the system aiming to prototype human-IoT experiences. 1 INTRODUCTION
The RelaxBreathe program was one of these prototyped and tested There is both clinical and consumer interest in remote sleep track-
human-IoT experiences. This positioning paper introduces Field ing using wearable technologies [10]. Sleep is complex and both
Lab Sleep and Energy, the RelaxBreathe program case study and objective and subjective, Sleep is personal and context dependent.
the findings of the team. Sleep varies over time. As such it is a suitable topic to be explored
by remotely tracking it at people’s homes. Liang et al.’s [7] paper
Copyright 2021 for this paper by its authors. Use permitted under Creative Commons describes SleepExplorer a visualization tool for personal sleep data
License Attribution 4.0 International (CC BY 4.0).
and contextual factors. Their sleep study was conducted for two
weeks in length. The authors emphasize the opportunity for being
able to provide tailored instructions and recommendations for be-
havior change. In order to do so a longer period of data collection
is necessary.
CHIIoT 1, February 17, 2021, Delft, The Netherlands Lovei, et al.
By applying the Data-enabled Design (DED) process [13] it is
possible to design prototypes for intelligent ecosystems. Using these
prototypes a multidisciplinary design (research) team collects ob-
jective and subjective data [8]. The built systems consist of (IoT)
products, services and people. The team is actively involved in
setting up, conducting, analyzing and communicating the results
of these studies. Studies in the past were conducted for a short to
medium period of time (1-4 months). However, it is not always
possible to learn enough about a certain domain, or about the ev-
eryday use of the designed intelligent ecosystem without deploying
it in the field for a longer period of time. Therefore, we decided to
explore how to conduct longitudinal design research for a period
of a full year focusing on the topic of remote sleep tracking using
IoT technologies. We built the Field Lab Sleep and Energy system.
In this positioning paper we are presenting a case study about how
to prototype Human-IoT experiences using Field Lab Sleep and
Energy.
Figure 2: The Field Lab system consisting of (1) a Withings
Sleep Analyzer [15], (2) a Somneo Wakeup Light [9], and (3)
the Field Lab Sleep mobile application used as a communi-
2 FIELD LAB SLEEP AND ENERGY cation platform. The arrows show the possibility of interac-
Field Lab Sleep and Energy is an intelligent ecosystem designed tion between the participants and the design probes.
for conducting longitudinal design research, exploring the topic of
remote sleep tracking using IoT technologies deployed in people’s 2.2 Remote data collection via the Field Lab
home environment. By applying the DED process [13] the design system
research team has made the following decisions for the design of Using the Field Lab system we can collect behavioral, experiential
the system. Field Lab Sleep and Energy consists of off-the shelf and contextual data. It is achieved by the deployment of the fol-
IoT devices that collect contextual, behavioral and experiential lowing design probes at participating families’ homes: (1) Withings
data related to remote sleep tracking and can be used in the home Sleep Analyzer [15], (2) a Somneo Wakeup Light [9], and (3) the
environment of its users. The built system is made to function for Field Lab Sleep iOS application.
a duration of a full year.
Using the system the team wanted to be able to prototype human- 2.2.1 Withings Sleep analyzer. The Withings Sleep Analyzer is a
IoT experiences based on the data collected via the remote sleep WiFi-enabled IoT device that is placed under the mattress in the
tracking activities. Therefore, we developed a way to be able to bed. The device collects data about its users’ sleep and sends the
gather, store, process, visualize and analyze the incoming data. collected data over WiFi to the Withings cloud. The Field Lab Sleep
Moreover, by building and embedding a communication platform and Energy system is able to retrieve the collected, and analyzed
in the Field Lab Sleep and Energy system the design research team sleep data from the Withings cloud via their official API [14].
could (1) instruct participants, (2) modify the functionality of the Data collected by Withings Sleep Analyzer that is used by Field
design probes, and (3) gather feedback about how the participants Lab:
are experiencing the ecosystem they’re using. • Time participant spends in bed (duration)
• Total sleep time (duration)
• Time awake (duration)
• Sleep efficiency (percentage: time asleep / total duration in
2.1 Conducted User Study bed)
• Sleep onset (timestamp)
From December 2018 till December 2019 we conducted the Field
• Sleep offset (timestamp)
Lab Sleep and Energy study. We deployed the Field Lab system in
the homes of five participating families from the Eindhoven region 2.2.2 Somneo Wakeup Light. The Somneo Wakeup Light is a WiFi-
of The Netherlands. The five participating couples were selected to enabled IoT device that uses light and sound to wake its users up.
be (1) healthy individuals with no (diagnosed) sleep disorder, and (2) It is possible to setup the alarm time and customize its sound via
tech savvy consumers. During the study we collected 2979 nights of the display of the light or via a smartphone application. There is
Withings data, 876 days of subjective sleep data, 109076 data points a relaxed breathing feature (RelaxBreathe) offered by the light as
about their bedroom environment, the participants opened the app well. This feature can be triggered manually via the display of the
4500 times, and 10350 Chat messages were exchanged via the built light. The light contains sensors that collect data about the bedroom
communication platform. The study was positively approved to be environment of the user.
conducted by the Internal Committee for Biomedical Experiments Data collected by Somneo Wakeup Light that is used by Field
(ICBE) of Philips. Lab:
Field Lab Sleep and Energy: A System for Longitudinal Remote Sleep Tracking and Prototyping CHIIoT 1, February 17, 2021, Delft, The Netherlands
• Temperature (degrees) 2.3.1 The back office of the Field Lab Sleep and Energy system. In
• Humidity (percentage) order to be able to analyze the incoming experiential, behavioral
• Sound level (decibel) and contextual data we built a back office that could be centrally
• Light level (lux) accessed by all the members of the (design) research team. The back
• Time RelaxBreathe feature was started (timestamps) office was deployed on AWS and data is stored in MongoDB. The
• Time RelaxBreathe feature was ended (timestamps) main functionalities of the back office are the following: (1) present
• Alarm usage (timestamps) data visualisations to the researchers based on the collected data,
(2) enable the researchers to play with the incoming data and the
2.2.3 Field Lab Sleep mobile application. The Field Lab Sleep mo- data visualizations.
bile application was custom-developed for Field Lab. It was de-
veloped using React Native [11] for iOS devices. First of all we
built a Chatbot to keep participants engaged during the study, ask
questions for gathering feedback, and for being able to remotely
instruct participants on how to use the devices. Secondly, we built
a Newsfeed to provide content to keep participants engaged, and
educate them about their sleep. Finally, we developed a way for
participants to look at visualized data reports and feedback. The
participants are encouraged to use the app on a daily basis and are
notified every time there is a new chat message, Newsfeed content
or personalized report sent to them.
Data collected by the Field Lab Sleep mobile application that is
used by Field Lab:
• App usage data (opening, closing timestamps, duration spent
on screens)
• Sleep related questions and answers: Sleep quality, Restful-
ness/Refreshed, Alertness, Consensus Sleep Diary [2]
• Qualitative questions and answers (scale from 1-10)
Figure 4: The Field Lab system allows the remote checking
and control of the the Somneo Wakeup Light
2.3.2 The communication platform of the Field Lab system. Using
the communication platform of the Field Lab Sleep and Energy
system the (design) research team can (1) schedule pre-written
Figure 3: Data visual showing the temperature data collected chat bot messages, and news feed articles, (2) compose reports
by the Somneo Wakeup Light [9] (top), the sleep data col- about the participants data that can be shown in the Field Lab Sleep
lected by Withings Sleep Analyzer [15] (bottom), and the mobile application. The chat bot messages are pre-written in the
subjective sleep quality as reported on a scale from 1-10 (best Flow.AI [3] platform that is used to define the logic of the chat
score is 10) by the study participant (right) bot. The back office of the Field Lab communicates with Flow.AI
via their WebSocket API [4]. The news feed articles are stored in
a MongoDB database. The researchers need to define their title,
subtitle, and a URL that points to the content to be shown to the
2.3 Components built for prototyping participants. The reports about the participants’ data are special
Human-IoT experiences using the Field Lab kind of news feed articles that contain a link to a PDF document
Sleep and Energy system that is stored in an AWS S3 bucket.
Using the Field Lab system the team was able to create prototypes
to (remotely) test Human-IoT experiences. This is achieved by us- 3 CASE STUDY: RELAXBREATHE PROGRAM
ing the following two components: (1) the back office and (2) the The Field Lab Sleep and Energy system was created in order to
communication platform. During the study we used the system to prototype Human-IoT experiences. Based on the collected data
modify the capabilities of the IoT components, tracked the usage of from the Withings Sleep Analyzer and the answers given to the
system components, instructed participants to use the devices dif- sleep related questions asked via the communication platform we
ferently, and gathered feedback about the experience of participants noticed that 5 of the 10 participants had troubles falling at least
using the system via the mobile application. twice a week. Therefore we decided to introduce the RelaxBreathe
CHIIoT 1, February 17, 2021, Delft, The Netherlands Lovei, et al.
program, a personalized, visually-guided breathing program that program has also shown the importance of improving the Field Lab
can prevent stress to accumulate during the day, in order to facilitate Sleep and Energy system to gather data about people’s motivation
sleep initiation and continuation at night. when testing out human-IoT experiences. Based on the suggestions
In order to prototype this experience the team has used the back of the users the program could be improved by further improving
office, the communication platform and the Somneo Wakeup Light. the connection to all the experiential, contextual, and behavioral
We developed a custom solution that uses a Raspberry Pi [6] with data that was collected from their homes.
Node-RED [5] installed on it to connect to the Somneo Wakeup light
via the local WiFi network. This way it is possible to read out sensor 4 CONCLUSION AND FUTURE WORK
data from the light and control the Relax Breathe feature. It was In this position paper we presented Field Lab Sleep and Energy.
achieved by sending AWS IoT [1] messages to the Raspberry Pi that The system was designed for longitudinal design research into
triggers the commands on the Somneo by first finding the device remote sleep tracking. The introduced system was used in a design
on the local WiFi based on its’ Mac address and then forwarding an research study for the duration of one year. We selected the IoT
HTTP POST command to the device over the local WiFi network. components that were deployed in people’s homes in a way that
The device itself is able to interpret the incoming message and based on the gathered, and analyzed remote sleep tracking data the
trigger the RelaxBreathe feature as if the participant pressed the team could come up and prototype new human-IoT experiences. By
respective button on its display. developing and embedding a communication platform the design
Using the communication platform we sent chat messages to research team was able to instruct participants to use the devices
each participant of the study that they could read inside the Field differently, track their device usage and gather their feedback about
Lab Sleep mobile application. By reading the messages they were the new way of using the IoT device. This way we could prototype
introduced to the program, and we asked whether they were inter- a new human-IoT experience.
ested in joining it. If they decided they wanted to participate they The presented case study shows how to remotely setup a re-
were guided to setup their schedule. Depending on their schedule laxation before sleep experience. The case study was successful
they received a Chat-based prompt together with a push notifica- for setting up and executing the program, and gathering feedback
tion to start the breathing exercises on their Somneo Wakeup Light. from the participants. Some participants have appreciated using it
Each participant could decide to (1) use the light for the breathing while others have provided valuable feedback to further improve
exercises, (2) delay the exercise for a later moment, or (3) to skip the the experience and the Field Lab Sleep and Energy system.
exercise. After they have finished with their exercise they got an During the one year the system was used for prototyping other
encouraging message from the system. If the participants decided experiences around the topic of family sleep [12], bedroom environ-
to do the breathing exercise the Flow.ai was programmed to ask the ment and sleep regularity. In all cases we used the same method of
back office to send a message to the Raspberry Pi via the AWS IoT instructing participants, potentially modifying the technical setup,
service to start the exercise over the local WiFi network to which tracking the usage interactions and gathering feedback using the
the Somneo light was connected. system. In the future we aim to explore other health technology
Next to the exercises we sent tips and news feed articles related topics using similar ecosystems that are designed by applying the
to the benefits of the relaxation before going to bed. After 7 days DED process.
the participants received a personal report that included a data
visual and had the opportunity to reflect on the program. They
could decide to (1) continue as before, (2) adopt their schedule, (3)
ACKNOWLEDGMENTS
or leave the program. The authors would like to thank Anne Wil Burghoorn, Melanie
Meyfroyt, Erwin Hoogerwoord, Thomas Visser and the rest of the
colleagues working on this project. We also would like to thank the
3.1 Results participants of the Field Lab Sleep and Energy study.
Two of the 10 participants have decided to continue the program
even though we have told them via the Chat messages that the pro- REFERENCES
gram was over. They reported that the program helps them falling [1] Amazon. 2020 (accessed January 4, 2021). AWS IoT. https://aws.amazon.com/iot/
[2] Colleen E Carney, Daniel J Buysse, Sonia Ancoli-Israel, Jack D Edinger, Andrew D
asleep better and helps them relaxing. We have also received feed- Krystal, Kenneth L Lichstein, and Charles M Morin. 2012. The consensus sleep
back on how to improve this experience by further personalizing diary: standardizing prospective sleep self-monitoring. Sleep 35, 2 (2012), 287–
the program to the user’s context. For one of the participants using 302.
[3] Flow.ai. 2020 (accessed January 4, 2021). Flow.ai. https://flow.ai
the light felt like a task instead of relaxation as this person did not [4] Flow.ai. 2020 (accessed January 4, 2021). Flow.ai Socket API. https://flow.ai/docs/
have problems with falling asleep. Other users have reported to api-docs/#socket-api
make the time of the exercise more easily customizable, as many [5] OpenJS Foundation. 2020 (accessed January 4, 2021). Node-RED. https://nodered.
org
times it happened that the exact moment they received the Chat [6] Raspberry Pi Foundation. 2020 (accessed January 4, 2021). Raspberry Pi. https:
prompt they were doing something else and then forgot about it. //www.raspberrypi.org/products/raspberry-pi-3-model-b/
[7] Zilu Liang, Bernd Ploderer, Wanyu Liu, Yukiko Nagata, James Bailey, Lars Ku-
Moreover, the participants have also provided feedback for how to lik, and Yuxuan Li. 2016. SleepExplorer: a visualization tool to make sense of
improve the IoT aspect of the program as they have asked us to correlations between personal sleep data and contextual factors. Personal and
introduce different channels or devices for delivering the exercise. Ubiquitous Computing 20, 6 (2016), 985–1000.
[8] Peter Lovei, Eva Deckers, Mathias Funk, and Stephan Wensveen. 2020. The Marios
One participant would have preferred voice based guidance instead and Luigis of Design: Design Plumbers Wanted!. In Companion Publication of the
of the light based one provided by our team. Last but not least, our 2020 ACM Designing Interactive Systems Conference. 197–201.
Field Lab Sleep and Energy: A System for Longitudinal Remote Sleep Tracking and Prototyping CHIIoT 1, February 17, 2021, Delft, The Netherlands
[9] Philips. 2019 (accessed March 10, 2020). SmartSleep Connected Sleep and Vol. 29. WILEY 111 RIVER ST, HOBOKEN 07030-5774, NJ USA, 231–232.
Wakeup Light. https://www.usa.philips.com/c-e/smartsleep/connected-sleep- [13] Janne van Kollenburg and Sander Bogers. 2019. Data-enabled design: a situated
and-wakeup-light.html design approach that uses data as creative material when designing for intelligent
[10] Anita Valanju Shelgikar, Patricia F Anderson, and Marc R Stephens. 2016. Sleep ecosystems. Ph.D. Dissertation.
tracking, wearable technology, and opportunities for research and clinical care. [14] Withings. 2020 (accessed January 4, 2021). Withings Developer Documentation.
Chest 150, 3 (2016), 732–743. https://developer.withings.com/oauth2/
[11] Facebook Open Source. 2020 (accessed January 4, 2021). React Native. https: [15] Withings. 2020 (accessed January 4, 2021). Withings Sleep Analyzer. https:
//reactnative.dev //www.withings.com/nl/en/sleep-analyzer
[12] T Vacaretu, S Pillen, AW Burghoorn, M Knufinke, P Lovei, S Overeem, C. van
derzwaluw, T Visser, and P Markopoulos. 2020. How did you sleep? Exploring by-
proxy sleep assessment in a field study setup. In JOURNAL OF SLEEP RESEARCH,