Forums

XXIII.1 January + February 2016
Page: 72
Digital Citation

From tracking to personal health


Authors:
Susanne Boll, Wilko Heuten, Jochen Meyer

Personal health has been gaining more and more attention lately, for good reasons. Two figures in particular provide strong motivation for this shift: In 2020, 73 percent of the world’s deaths will stem from noncommunicable diseases [1]. These are largely preventable [2], particularly by reducing risk factors such as smoking, poor nutrition, and inadequate physical inactivity. Hence the research field of personal health has become very important. Its defining characteristic is that the individual plays an important role in transitioning to and maintaining a healthy lifestyle.

Insights

ins01.gif

The miniaturization of sensors and actuators as well as new battery technology have enabled the development of mobile apps and wearable devices that can detect a plethora of health-related data, such as steps taken, miles cycled or run, heart rate, sleep quality, and muscular effort. The hidden marketing promise is that merely tracking your health will make you healthier. However, tracking is only the beginning of a long pathway toward a healthier lifestyle, from monitoring health to taking action.

The first barrier is the actual device itself and its use in daily life. Today, many devices are bought as a kind of toy, and to this extent, they are ends in themselves. For tracking to become successful, however, the device must be seamlessly integrated into daily life. But many people do not like technical devices—they forget to wear them and to charge the batteries; they find they do not match their personal taste; or, in certain situations, they are not even allowed to wear them.

Another barrier is understanding and interpreting the collected data. Step counting seems to have become the silver bullet of activity tracking, and—though not an official recommendation—walking at least 10,000 steps a day has become a primary fitness goal. But what does this mean? Do these 10,000 steps imply that I am healthy? If I miss out on this goal, am I unhealthy? Are 10,000 steps good for everyone? How are non-stepping activities such as cycling accounted for? Tracking devices keep us focused on low-level data rather than on higher-level personal health goals. Simple counting fails to answer questions such as: Do I live in a healthy way? What can I do to achieve a better lifestyle? Many people have health goals, either explicit or implicit, as part of their daily routine. And these health goals, rather than steps, counts, and scores, must become the benchmark for personal health.

Experiences from our own [3] and related recent work [4] show the many obstacles that individuals face when seeking support for a healthier lifestyle. Perception and appearance, wearability, intensity of interaction, and activity measures and validity of data together play an important role. For tracking to be successful in promoting personal health, we will need designs, concepts, and data models that match the user’s personal preferences and intentions.

Plurality of Devices and Data Ownership

Another key obstacle is the siloing of data between devices and manufacturers. If we could easily switch from one device to another, individuals could choose when and how to use—or not to use—any given device at any point in time. This would address the demand for multiple devices to work for the same health feature, and for observing multiple parameters using different devices—the small and simple activity tracker that is acceptable during work hours, but the cool-looking one with lots of features worn for leisure time; the wristband worn for activity tracking, the ambient bed sensor for sleep monitoring. Consequently, mobile phones or smart watches will surely be important building blocks of personal health-monitoring systems. They may become the information terminal, the universal journaling system, or the fallback device for general behavior sensing. But we do not believe that either will ever be the one and only device. Individuals are selecting and swapping different devices for different reasons, and these devices change over time as new products enter the market.

With this in mind, the integration of health-related data from multiple sources and providers is a key requirement for future health systems. While many manufacturers and portal providers integrate third-party sources into their services, they often are quite restrictive in granting others access to their own data. APIs may not exist at all, and if they do there may be restrictions regarding which and how much data can be accessed. A simple export as a .csv or an Excel file for the user’s own records is often impossible. Fitbit even recently introduced an artificial 30-day limit for manually exporting data, foiling its claim of “your data belongs to you.” Still, we remain optimistic that market developments, user demands, or regulatory interventions will ensure truly free access to one’s own data, enabling exciting new opportunities for user-oriented health applications. Systems such as TicTrac (Figure 1; https://www.tictrac.com/) or Zenobase (https://zenobase.com/) already do a nice job in integrating multiple sources. They access portals such as Fitbit, Garmin Connect, or MyFitnessPal to gather a user’s activity, nutrition, and other data, and evaluate the data against given goals. But these systems tend to be difficult to use and focus on the analytical presentation of data with bar and pie charts rather than on intuitive visualizations. The systems today address the data-enthusiast quantified-selfer much more than the moderately engaged Jane and John Doe who do not want to become data scientists of their own health data.

From Scattered Data to a Personal Health View

The use of different tracking devices leaves the individual with a patchwork of health-related data. However, looking at individual and unrelated health features is not sufficient. It is well understood that nutrition, sleep, physical exercise, mental state, and numerous other non-clinical factors all contribute to a healthy lifestyle. While single tracking devices exist for individual health-related features, we lack methods and tools that help individuals observe and understand their personal health in its entirety. In our Lotus system, we fuse and analyze a user’s data from different activity trackers, fitness portals, networked scales, and sleep monitors into one data portal. The portal not only collects the different data sources under one roof but also aims at the aggregation, fusion, and interpretation of the data (Figure 2). With an architecture that aggregates low-level health features such as steps or calories burned into higher-level personal health values and scores, we aim to provide a holistic view of an individual’s heart-friendliness behavior. In a long-term study, we are trying to understand whether personal reflection on this holistic view of the user’s health changes the user’s attitude toward health.

Finding Opportune Moments for Health

Mobile phones and smart watches have great potential for digital interventions, helping the individual to make changes or to stay on track for a personal health goal. However, when to offer such digital interventions still needs to be understood. With their comprehensive sensor technology, wearable devices offer the opportunity to detect individual behavior and context and, from this, to anticipate opportune moments for delivering health-related messages to the individual. Recent research looks at the rationales for when individuals would be receptive to messages during the day on their mobile devices. To understand these opportune moments for health interventions, we investigated the contexts in which users typically perceive and react to (or ignore) mobile notifications [5]. We analyzed the usage of several mobile phone sensors to see whether they could help us understand when people respond to notifications. Based on observations of when individuals respond, we can now issue notifications at times and in contexts when the notification has a higher chance of being attended to. This creates the opportunity to remind individuals about their health goals in moments when they might be available for the message and potentially for the health-related action.


Many systems today take a perspective of days, weeks, and months. But to support individuals on their lifelong journey to personal health, we also must take a lifetime or life span perspective.


Time for Digital Interventions—from Moments to Life Spans

The temporal perspective of when can range from an actual moment to longer time spans. The shorter perspective is about moments on this very day or this recent week, and involves encouragement to achieve goals, and often to induce behavior changes. For different reasons, many technical systems today take a perspective of days, weeks, and maybe months. But to support individuals on their lifelong journey to personal health, we also must take a lifetime or life span perspective. This is about times when one’s life changes, about self-understanding, and ideally about making informed decisions about health behavior. Long-term monitoring is feasible with today’s technology (provided we take into account the user requirements on sensor use, as described earlier). Our vision is of the personal health system as a lifelong “navigation system for health” [6,7]. Such a system would remain in the background over years of relative stability but become active when one’s life changes and some adjustment is required. Examples might be life events that require behavior changes to sustain a healthy state, such as having children, changing jobs, or moving to another city. The system might also detect slow changes that can be hard to notice personally, such as the gradual increase in weight by one pound each year until someone moves out of their target weight range. We need to move from thinking about astronomical (day, month, year), cultural (week), and mathematical (decade) time spans to considering personally significant time spans and events when designing health-support systems.

The Last Mile—Delivering Wearable Health Interventions

The last step of delivering digital interventions on mobile phones, smart watches, or other wearable devices is the actual message and presentation design. For a digital intervention to be successful, the interface design is crucial. What we call the last mile is the question of how to deliver an intervention such as a reminder to be physically active or a warning to avoid a situation related to smoking. While text messages and notifications are the classic means for this, it is not yet understood what message on which device in which presentation format works best. Recent work has described this as a theoretical gap that must be bridged between the behavioral sciences and human-computer interaction research [8]. A message needs to trigger the right action; therefore, we need to carefully design messages that motivate us, cheer us up, or warn us in the right way. The device and the modality of the message are also highly important. Messages that beep at the wrong time may be annoying; quiet notifications may not be perceived. Adaptation to the individual’s context and personal preferences are crucial. We will also need to better explore what role aesthetics and fashion will play in delivering mobile health interventions. Smartphones and smart watches will be two out of many devices in the future. Will fashion accessories be more successful? In recent work, we designed and evaluated a custom-made smart bracelet for supporting regular water intake, WaterJewel [9]. The peripheral display was more effective than a comparable app on a smartphone and the aesthetics and form factor—important user requirements for wearable health devices [10]—were also appealing. While these prototypes and studies are only examples, they show the importance of understanding how to deliver appealing health interventions to individuals.

Conclusion

There are many barriers and obstacles to personal health, from an individual’s actions to the actual devices used to monitor those actions. Only by addressing all these different research challenges, from measuring and understanding the individual’s personal health needs to tailoring digital interventions to the individual’s personal context in daily life, will we be on the right path to technology support for personal health.

Acknowledgments

The discussion of the potentials and challenges of personal health technologies has been inspired by many wonderful researchers, not all of whom we can cite here. This includes, but is not restricted to, the magnificent participants of a recent Dagstuhl workshop on “Life-long health behavior-change technologies” (http://www.dagstuhl.de/de/programm/kalender/semhp/?semnr=15262), as well as recent workshops at CHI, CSCW, and PervasiveHealth. We look forward to the developments to come in this exciting field of work.

References

1. 2008–2013 Action Plan for the Global Strategy for the Prevention and Control of Noncommunicable Diseases; http://www.who.int/nmh/Actionplan-PC-NCD-2008.pdf

2. Global Action Plan for the Prevention and Control of NCDs 2013–2020; http://apps.who.int/iris/bitstream/10665/94384/1/9789241506236_eng.pdf

3. Meyer, J., Fortmann, J., Wasmann, M., and Heuten, W. Making lifelogging usable: Design guidelines for activity trackers. Multimedia Modeling. Springer, 2015.

4. Epstein, D.A., Ping, A., Fogarty, J., and Munson, S.A. A lived informatics model of personal informatics. Proc. of the 2015 ACM International Joint Conference on Pervasive and Ubiquitous Computing. ACM, New York, 2015, 731–742.

5. Poppinga, B., Heuten, W., and Boll, S. Sensor-based identification of opportune moments to trigger unobtrusive notifications. IEEE Pervasive Computing 13, 1 (2014), 22–29.

6. Meyer, J., Lee, Y.S., Siek, K., Boll, S., Mayora, O., and Röcker, C. Wellness interventions and HCI: Theory, practice, and technology. ACM SIGHIT Record 2, 2 (2012), 51–53. DOI:10.1145/2384556.2384564

7. Meyer, J., Simske, S., Siek, K.A., Gurrin, C.G., and Hermens, H. Beyond quantified self: Data for wellbeing. CHI Extended Abstracts. ACM, New York, 2014, 95–98. DOI:10.1145/2559206.2560469

8. Hekler, E.B., Klasnja, P., Froehlich, J.E., and Buman, M.P. Mind the theoretical gap: Interpreting, using, and developing behavioral theory in HCI research. Proc. of the SIGCHI Conference on Human Factors in Computing Systems. ACM, New York, 2013, 3307–3316.

9. Fortmann, F., Cobus, V., Heuten, W., and Boll, S. Waterjewel: Design and evaluation of a bracelet to promote a better drinking behaviour. Proc. of the 13th International Conference on Mobile and Ubiquitous Multimedia. 2014, 58–67.

10. Fortmann, F., Heuten, W., and Boll, S. User requirements for digital jewellery. Proc. of the 2015 British HCI Conference. ACM, New York, 2015, 119–125.

Authors

Susanne Boll is professor of media informatics and multimedia systems in the Department of Computer Science at the University of Oldenburg, Germany. Her research interests lie in the fields of semantic retrieval of digital media, mobile and pervasive systems, and intelligent user interfaces. Her recent focus is on pervasive interaction for personal health systems. susanne.boll@informatik.uni-oldenburg.de

Wilko Heuten is senior principle scientist and manager of the Interactive Systems group at OFFIS – Institute for Information Technology in Oldenburg, Germany. His research interest is the design and development of pervasive and multimodal interactive technologies to support everyday life activities. wilko.heuten@offis.de

Jochen Meyer is the director of the R&D division Health in the OFFIS Institute for Information Technology in Oldenburg, Germany. His research interests lie in technologies for prevention and well-being, ambient assisted living, and personal use of multimedia data. meyer@offis.de

Figures

F1Figure 1. TicTrac. Lots of great features, but maybe challenging to understand.

F2Figure 2. Aggregation to turn data into meaningful information.

©2016 ACM  1072-5220/16/01  $15.00

Permission to make digital or hard copies of all or part of this work for personal or classroom use is granted without fee provided that copies are not made or distributed for profit or commercial advantage and that copies bear this notice and the full citation on the first page. To copy otherwise, to republish, to post on servers or to redistribute to lists, requires prior specific permission and/or a fee.

The Digital Library is published by the Association for Computing Machinery. Copyright © 2016 ACM, Inc.

Post Comment


No Comments Found