Forum

XVIII.3 May + June 2011
Page: 22
Digital Citation

IT in healthcare


Authors:
Elizabeth Mynatt

Healthcare is both deeply personal and staggeringly complex. It starts with individual physical bodies and stretches to include an array of technologies, physical places, caregiver networks (both formal and informal), and decision-making tasks. Each configuration of body, people, technology, places, and decisions presents compelling and timely challenges for human-centered computing.

As the introductory article for this forum, my aim is to describe this landscape of challenges for our community and to invite articles on these topics.

Health Is About Sensing

In contrast to other areas of HCI, which focus on people producing digital objects, working across electronic networks, and creating virtual economies, health is fundamentally rooted in the physical world, due to its undeniable connection to the human body. With the foundation of the physical body, and branching out to the physical places that embody healthcare, current and future research seeks to sense information in this physical world and make it available for analysis and action. Sensors on or near the body attempt to assess physiological state; capture actions directly related to the physical state, such as what someone is eating; recognize key activities, from walking to taking medication; and integrate data about actions over time.

As sensing provides the link to the physical world, the need to develop reliable, robust, efficient, and acceptable sensors remains a constant challenge, especially in situations requiring around-the-clock monitoring in places outside of healthcare institutions, such as the home. Sensors must perform in difficult and dynamic physical conditions. Their signals must be reasonably reliable and amenable to effective analysis. Many sensors strive for 24/7 operation over long periods of time, hence the need for energy-efficient and energy-harvesting designs.

Sensing technology also needs to be deemed acceptable by various stakeholders. Objections to sensors can range from the sensors being physically uncomfortable to visible sensors creating social stigma. Invasive sensing in a hospital setting does not transfer to typical home environments. The relationship between sensing and privacy raises many thorny challenges that ultimately involve the use of sensed information and the elasticity of privacy concerns in the face of specific healthcare challenges. However, the base capabilities of a sensor—the capabilities of video cameras in contrast to motion sensors, as well as single-latch sensors—set the stage for privacy considerations. Designers must often lean toward minimally invasive sensing, despite the overall utility of more general-purpose sensors.

While its connection to the physical world is fundamental, overall health is a combination of physical, mental, and social well-being. This “health triangle” makes it clear that health extends beyond the physical body and physical injury. Enabling mental and social well-being draws on many competencies in HCI, ranging from persuasive computing to online communities, as discussed further in this article.

Health Is About Data

Bill Stead, associate vice chancellor for strategy/transformation and chief information officer at Vanderbilt University Medical Center, spoke last year at the national meeting Discovery and Innovation in Health IT [1]. His assessment was that physicians were already faced with an overabundance of data, yet they are only at the beginning of the eventual tsunami of data coming within the next decade. Any challenge seen in data-visualization research is present in healthcare data—for example, time varying, multivariate, heterogeneous, noisy data stemming from disparate sources and utilized by, at a minimum, a host of professional healthcare providers. Additionally, data analysis is not static but needs to convey emergent phenomena. How can someone roll back time to interrogate an older visualization in the face of new data? Information-presentation techniques must also accommodate a spectrum of people who have varying levels of analytical skills and available time.

A recurring challenge in data visualization analysis is the detection and portrayal of trends. Many forms of healthcare assessments are based on identifying declines or improvements in people’s abilities and conditions. For example, declines in physical locomotion, i.e., walking, could portend a damaging fall or could indicate the onset of disease. Detecting trends in cognitive abilities is even more challenging, as cognitive performance varies considerably due to many external and internal factors. Even the task of determining if a behavior is “unchanging” or within normal parameters is challenging, as the definition of “normal” is frequently patient-specific.

The reliance on sophisticated data analysis for healthcare concerns will draw the HCI and machine-learning communities even closer. That said, the technical capabilities to produce various forms of data analysis far exceed the human capacity and desire to attend to this information. The use of healthcare data is always embedded in constrained human environments: 15-minute physician visits, nursing shift changes, time-limited surgical procedures, busy and distracted caregivers, and simply people who would rather focus on the events of the day in contrast to monitoring and reflecting on their own health.

Health Is About Decisions

Professional caregivers rely on a foundation of education and training; individual patients rely on a foundation of customs, habits, and daily routines. From determining a diagnosis and course of treatment to deciding what to eat for lunch, the stakeholders in healthcare attempt to integrate new information in the course of routine to emergency decision making.

Integrating new information is itself a decision-making task. The search for online information brings this challenge to the forefront: how to compose a search query and then follow a thread of links while trying to determine if the advertised (not just in the commercial sense of the word) information is relevant and reliable? Frequently, the next step is to collaborate with other stakeholders, from caregivers to physicians, to assess the information and determine its implications.

The road from data to diagnosis has received considerable attention in the medical and computing communities, from imaging technologies to AI expert systems. More recently, greater attention has been paid to the process of healthcare delivery and the iterative decision-making processes it comprises. Management of many chronic diseases (diabetes, asthma, heart disease, etc.) relies on iteratively calibrating treatment factors, including medications and daily behaviors. For example, an asthmatic patient or family caregiver must integrate the use of daily and rescue medications with respect to dynamic changes in behavior and environmental conditions. That patient’s physician should assess ongoing treatment plans in terms of whether the patient has returned for a routine visit or is in the emergency room.

This increasing load on patients has motivated a new generation of computing tools that enable patients to better manage chronic conditions. With a thin educational foundation, patients make decisions every day, whether implicitly or explicitly, on how to adhere, or not adhere, to a prescribed treatment plan in the face of their own behavior. These tools must acknowledge that patients will rarely be 100 percent compliant in moderating human behavior and yet provide the scaffolding so that patients can be more successful in improving health outcomes. These interfaces frequently motivate patients into becoming “detectives” in making sense of their own health data, as well as helping patients set localized goals that should lead to better health [2]

As patient behavior is more critical to the prevention and management of modern-day chronic diseases, there is growing interest in designing computing experiences that influence healthy behavior. These persuasive experiences come in many forms, relying on play and entertainment and tapping into social support and competition. Trade-offs and combinations of interfaces that emphasize management and goal setting with interfaces that compel patients to “do the right thing” create a large design space for human-centered computing. While systems that rely on increasing amounts of personal data raise privacy concerns, persuasive systems that emphasize motivation over introspection also raise ethical concerns.

Health Is About People

Despite my emphasis on patient and providers, the reality is, of course, that health is about people. Recent emphasis on “wellness informatics” makes this point with more clarity [3]. Good health is a universal desire, and healthcare does not begin in the doctor’s office with a disease diagnosis. Health is intricately embedded in daily life and the people who populate daily experiences.

Research by Consolvo and others identified the “care network” as a critical foundation for designing health-informatics applications and services oriented to people in their homes and those who take care of them [4]. This perspective is almost universal in healthcare. Although health is grounded in the body, there is almost always a network of care made up of the individual; his or her family, friends, neighbors, and colleagues; and, of course, another network of more formal healthcare providers. How this network shares information, coordinates care, builds trust (or not), and shares differing perspectives and expertise is the foundation for healthcare in everyday settings.

One critique of research in healthcare informatics is the focus on disease. Partly a result of funding agencies, as well as an overall orientation to intervention as opposed to prevention, this perspective flies in the face of everyday experiences of celebration and life satisfaction. Although people share varying degrees of aversion to risk and a desire to be healthy, daily choices regarding food, sports, sleep, social engagement, and so on are generally not viewed first through the lens of disease. The strong influence of social networks has made some in the medical community describe chronic diseases such as diabetes and heart disease as contagious. Patterns of daily behavior are profoundly shaped by the people who jointly participate in these activities. These social networks bring their own desires for an enjoyable life.

Social networks are no longer limited to face-to-face interactions. They now span the globe in many forms ranging from online communities, shared interest and support groups, and digital hubs such as Facebook—areas ripe with potential. Understanding how online social networks can support a wide range of information seeking, decisions, and social-support needs in healthcare is important. While shared-interest and support groups have been around since the very first news groups, online sites such as PatientsLikeMe.com are causing a stir in the medical community. Participants share their personal medical data in the hopes of identifying alternative treatment options, with the goal of establishing scientific evidence around the prognosis and treatment of some of the most vexing diseases in modern life.

The goal of this article is to broadly depict topics in personal-health informatics that are of interest to the HCI community. In emphasizing personal or patient-centered healthcare, I have not addressed other major areas of inquiry related to public health and medical discovery. One reason for optimism about the role of human-centered computing in healthcare research is the growing recognition of the repeated need for HCI innovation in healthcare technologies. Both the recent CCC report [5] and the U.S. PCAST report on healthcare information technology [6] call for effective interfaces for providers and patients for improving healthcare outcomes. My hope is that this forum will provide fertile ground for the HCI community in tackling the many challenges in personal-healthcare informatics.

References

1. Computing Community Consortium, Discovery and innovation in health IT, 2009; http://www.cra.org/ccc/healthit.php

2. Mamykina, L., Mynatt, E., Davidson, P., and Greenblatt, D. MAHI: Investigation of social scaffolding for reflective thinking in diabetes management. Proc. of the 26th Annual SIGCHI Conference on Human Factors in Computing Systems. (Florence, Italy April 5-10). ACM, New York, 2008, 477–486.

3. Grinter, R.E., Siek, K. A., and Grimes, A. Wellness informatics: Towards a definition and grand challenges. Proc. of the 28th of the International Conference Extended Abstracts on Human Factors In Computing Systems. (Atlanta, GA April 10-15). ACM, New York, 2010, 4505–4508

4. Consolvo, S., Roessler, P., and Shelton, B. E. Computer-supported coordinated care: Using technology to help care for elders. IEEE Pervasive Computing 3, 2 (2004), 22–29.

5. Graham, S., Estrin, D., Horvitz, E., Kohane, I., Mynatt, E., and Sim, I. Information technology research challenges for healthcare: From discovery to delivery. Computing Community Consortium Whitepaper Series. Computing Research Association (CRA), May 2010.

6. President’s Council of Advisors on Science and Technology (PCAST). Realizing the Full Potential of Health Information Technology To Improve Healthcare for Americans: The Path Forward. Executive Office of the President. 2010.

Author

Elizabeth D. Mynatt, professor of interactive computing at Georgia Tech, is the executive crector of the Institute for People and Technology (IPaT). A SIGCHI Academy member, Mynatt’s research in everyday computing investigates personal health informatics, domestic computing, and computer-supported collaboration.

Footnotes

DOI: http://doi.acm.org/10.1145/1962438.1962446

Sidebar: Sensing

FitBit http://www.fitbit.com An unobtrusive accelerometer-based sensor that clips to a belt or pocket. Reports steps, calories burned, and sleep quality. Syncs to web service and social media.

BodyBugg http://www.bodybugg.com Multi-sensor armband that tracks physical activity and reports directly to a smartphone.

Adidas MiCoach http://www.adidas.com/us/micoach/ An accelerometer-based pedometer that works with a personal training web service.

Philips DirectLife http://www.directlife.philips.com/ An accelerometer-based pedometer that works with a personal training web service.

Ant+ Network http://www.thisisant.com/ant/ant-interoperability Wireless networking technology used by many sensors in health/wellness (e.g. heart rate watches, Adidas MiCoach). Useful because of its low power consumption and relative ubiquity.

Nike+ http://nikerunning.nike.com/nikeos/p/nike-plus/en_US/ Shoe-mounted pedometer integrated with iPods/iPhone. Combined with social network features that let runners compare runs and receive real-time social support via Facebook during their run.

WiThings Body Scale http://www.withings.com/en/bodyscal Wifi-enabled bathroom scale that connects to a web service or iPhone/iPod and lets users visualize their weight over time and share with others. Auto-detects users in multi-person households.

Sidebar: Data

Patients Like Me http://www.patientslikeme.com Web platform that enables people to share information that can improve the lives of patients diagnosed with life-changing diseases.

Sidebar: Decisions

Health Month; http://healthmonth.com/ Social resolutions game. Adds a gamification layer and social persuasion techniques to everyday commitments. Wellness-centered.

Sidebar: People

Online Health Communities (beyond Patients Like Me)

SparkPeople http://www.sparkpeople.com Weight loss community

dLife http://www.dlife.com Diabetes community

FatSecret http://www.fatsecret.com Weight loss community

©2011 ACM  1072-5220/11/0500  $10.00

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 © 2011 ACM, Inc.