Morning Session

Dr. Robert Croyle, Director of the Division of Cancer Control and Population Sciences (DCCPS), National Cancer Institute (NCI), welcomed participants. He noted that this meeting is a followup to a meeting held in September 2003 in South Carolina. That meeting provided a general overview of real-time data capture; Dr. Croyle explained that the Capturing Physical Activity and Diet in Real-Time Meeting would focus on areas of application and behavior. The NCI is trying to improve its efforts as a funder of behavioral science research with regard to supporting initiatives related to methodology and measurement. The intent is to diffuse and disseminate awareness and interest in a variety of new methods and explore the public health applications of behavioral science in greater depth. He noted recent growing epidemiological evidence of a link between obesity, weight gain, body mass index, and cancer etiology. As a result, the NCI is expanding its efforts in the area of energy balance and cancer as a new chapter in the Bypass Budget. A new NCI effort is being undertaken to fund P50 Centers on energy balance and cancer. In addition, the NCI and National Institutes of Health (NIH) staff have worked to incorporate research methods and methodology into the NIH Roadmap. Some of the initiatives resulting from these efforts include language that addresses topics related to this meeting. Dr. Croyle encouraged participants to use this meeting as an opportunity to provide ideas and suggestions regarding NCI's prioritizing issues and areas of importance.

Dr. Audie Atienza, meeting Chair and Program Director in the Health Promotion Research Branch, Behavioral Research Program, DCCPS, NCI, welcomed participants and discussed the importance of focusing on physical activity and diet research. He noted that 60 percent of adults in the United States currently are considered overweight or obese, and one-third of adults are sedentary and inactive. The September 2003 meeting held in South Carolina brought together researchers in various areas to promote greater understanding of the behavior of individuals in the natural environment. Dietary assessment and physical activity emerged as areas in which there is a great need to understand reasons for behavior in the natural (versus laboratory) environment, and to develop interventions aimed at changing behavior in real time. The agenda for this meeting was developed with this need in mind, as well as to provide information on assessments, interventions, and technological advancements and solicit ideas from the extramural community regarding what is critical and feasible for future research in these areas.

Dr. Saul Shiffman, Professor of Psychology at the University of Pittsburgh, stated that he and Dr. Arthur Stone coined the term Ecological Momentary Assessment (EMA). The "ecological" aspect refers to studying people in their natural environments in real-world experience, where some phenomena are best studied. The "momentary" aspect is intended to avoid recall and to allow people to report current experiences. The "assessment" component includes self-reports, measures of performance, and instrumented measurements that do not require the subjects' participation.

The rationale EMA and its use of diaries include concern about recall bias; interest in real-world experience and ecological validity; interest in associations between environment, experiences, and outcome; and interest in process over time. With regard to recall bias, it has been found that retrospective recall often involves reconstruction, which may be inaccurate and/or biased. The use of diaries instead of recall allows researchers to collect data in the real world, although quality concerns do arise. EMA methodologies can be used to answer both between-subject (aggregated data) and within-subject (disaggregated data, often over time) types of questions. Sampling in EMA refers to sampling "moments" or occasions within the life of a person. Sampling can involve events (e.g., discrete episodes of vigorous exercise or food consumption) or can be time based (i.e., fixed times, as in completing a diary at specified or variable times, as in random sampling). Multiple sampling techniques can be incorporated into a single study design. Dr. Shiffman emphasized the volatility of life and the need to design studies to capture this volatility.

In discussion, Dr. Karen Glanz, Professor of Behavioral Sciences and Health Education at the Rollins School of Public Health, Emory University, noted that the issue of forward filling (making diary entries in advance) underscores the fine line between self-assessment and monitoring, and that forward filling actually may reinforce behavior. Dr. Shiffman noted that perhaps the key difference is between what subjects plan to do and what they say they have done. Dr. Lenore Arab, Senior Director and Head of Global Epidemiology at Amgen, Inc., explained that writing rather than dictating diary entries may force subjects to be more aware of what they are doing because they have a written record of their behavior. Dr. Shiffman characterized this issue as an open question that needs to be addressed.

Dr. Paul Cinciripini, Professor of Behavioral Science at The University of Texas M.D. Anderson Cancer Center, University of Texas, addressed the issue of translating data into interventions that might be used to change people's behavior. Health interventions designed to improve health and well-being include both lifestyle changes and quality-of-life changes. It is important to note that the health outcome may be distal to the intervention, and the immediate health consequences may not be readily apparent. Challenges to implementing effective health interventions include compliance, maintaining interest and motivation, and dynamic problem solving (e.g., the counselor or provider typically is not "in the moment" with the patient when a problem with the intervention may arise and must provide advice that the patient may apply the next time a problem is encountered). Hand-held interventions can help address these challenges by including self-reports and objective assessments done in real time, and by helping to identify antecedents and consequences more easily so that corrective measures may be applied early. Hand-held computers are easy to carry and use; have multiple functions; deliver significant amounts of information on demand; can prompt users for interaction; can employ complex, interactive programming to implement planned interventions in real time; and can be integrated with other electronic forms of assessment or physiological monitoring devices.

Momentary health interventions may be conceptualized along three dimensions:

• Interventions that differ in their primary approach to behavior change (e.g., education, monitoring and feedback, delivery of specific intervention strategies to increase or decrease target behavior)
• Interventions that differ in user- versus device-initiated assessments
• Interventions that differ in complexity of the algorithms used to deliver intervention content.

Simplifying such interventions and making them easy to use is a key challenge. Devices may be used alone or in combination with other intervention techniques, such as telephone counseling. A major goal is to deliver an intervention at the precise moment when it will impact an individual's behavior. Maintaining motivation and proper programming are other key factors in the success of such techniques.

In discussion, Dr. Abby King, Professor of Health Research, Policy, and Medicine at Stanford University School of Medicine, emphasized the need to work toward applying techniques geared to promoting individual behavior change in a population context to address population-based epidemics. Dr. Barbara Ainsworth, Professor and Chair, Department of Exercise and Nutritional Sciences at San Diego State University, noted that many challenges are involved in these activities. Dr. Shiffman commented that the laws of psychometrics and sound measurement practice apply to programs that use hand-held devices just as they apply to other kinds of programs. Dr. Cinciripini mentioned the notion of using incentives to promote compliance among those involved in studies. Dr. Ruth Patterson, Program Director of Health and Human Services at The San Diego Foundation, noted that although random errors may exist in the underlying database, the database still can be useful.

Dr. Stephen Intille, Technology Director of the Changing Places/House_n Project in the Department of Architecture, Massachusetts Institute of Technology, stated his belief that technology has its greatest potential in scaleable intervention on a personal level. Technology will allow interventions to be delivered to millions of people at low cost. These changes should be seen in 3-5 years. Technology also has huge potential to establish intervention and communication links between people. Currently, it is easier to apply technology to physical activity than to dietary assessment. Dr. Intille provided brief demonstrations of a few of these technologies.

Dr. Intille indicated that new technology now is enabling: continuous, rich recording of data from a variety of sensors; algorithms to process such data and reduce coding time for researchers; context-sensitive data collection to collect data and prompt for self-reports at desired times and places; and context-sensitive, personalized interventions. This will allow health researchers to better study how context impacts behavior and enable them to create and measure the impact of "just-in-time" interventions. He urged researchers to let device manufacturers know specific applications they would like to see made possible.

In discussion, Dr. Arab asked what the range of the sensors is and how sensors might distinguish between, for example, different family members in a household. Dr. Intille responded that automatic detection becomes more difficult when more than one person is in a particular environment. Researchers are addressing this problem. Range is not a major problem, and individual personal digital assistants (PDAs) can be used when more than one person is to be monitored. Dr. Glanz noted that sometimes the people who most need an intervention are the ones who are least likely to use it. Dr. Intille responded that small incremental changes (e.g., scanning two bags of chips and choosing the healthier one) are valuable and can lead an individual to further change in the future.

Afternoon Session

In the afternoon session of the meeting, attendees participated in two small working group sessions. The first small working group session consisted of two groups, one focused on quantitative/qualitative assessment and methodology (the Assessment Group), and another focused on behavioral interventions and the issues faced when conducting trials, from multiple levels (the Intervention Group). Both groups were asked to discuss the following questions during the first small working group session:

• What are the five most promising opportunities relevant to the use of real-time methodology and related technology in physical activity and/or diet assessment (or intervention)?



• How can real-time methodology address issues that cannot otherwise be answered with current (not real time) methodologies?



• What are five main barriers/challenges in real-time physical activity and/or diet research?
  • How are these barriers/challenges similar or different from general (not real-time) physical activity and diet research?
  • What are the issues relevant to special populations such as youth, older adults, low literacy, ethnic minorities, and those with disabilities?
  • What has kept most researchers from incorporating newer technology/methodology into their studies?


• What strategies and resources are needed to further incorporate real-time methodology and related technology in physical activity and/or diet research?

Dr. Glanz served as spokesperson for the Intervention Group, noting that discussions emphasized creating a bridge between the individual and environmental levels. The group discussed setting- or context-based feedback, including interventions that are automated at the point of choice such as scanners, global positioning system (GPS), and sensor messages. For example, in the grocery store or cafeteria, kiosks might help guide people toward particular or healthier choices, and "smart cards" could be used to help record choices and provide feedback.

Another area discussed was that of combining assessment and individual cues, such as PDAs used to cue people to exercise, along with the possibility of using triggering events such as a person sitting for too long, opening the refrigerator too often, and so forth. The area of commercialization was somewhat controversial, but group members agreed that tying in and partnering with for-profit marketers are ways to increase incentives. Examples include coupons that result from behavioral cues from real-time database interventions and individual and environmental adaptations of gym equipment with activity sensors. Similar approaches might be used in other types of physical-activity environments, including park trails and walking areas, and perhaps could combine technology with people wearing devices such as step counters. Cellular telephones and wireless technologies are additional examples of possible combinations of social and feedback types of interventions. A recurrent theme of the discussions was a focus on more passive data collection that does not create a burden for collectors or users.

Dr. Glanz noted that group members saw the question of how real-time methodologies might address issues that cannot otherwise be answered with current (not real-time) methodologies as somewhat of a "gray" area in that there are tasks that can be accomplished particularly well via real-time technologies, but other technologies also may be capable of accomplishing these tasks. Examples of such interventions include:

• Point-of-choice real-time decision aids that can be applied when people are in the midst of a decisionmaking process.
• Unobtrusive monitoring of what people are doing can help to avoid recall problems.
• Decision-free devices can sense what people are doing versus having people report their activities in real time.
• Feedback can be made more relevant by immediacy, and can allow individuals to experience success if they do what they have been advised to do.
• People can link directly with information sources and telemedicine.
• Artificial intelligence can be used to substitute for human counseling (as in Dr. Ken Resnicow's work in which motivational counseling may be accomplished by artificial intelligence using PDAs), and can focus less on informational feedback and prepackaged motivation and more on motivational interviews to draw out individuals' motivations.
• People can enhance social networking and capture environmental context.
• Portability-not having something stationed either on a desktop computer or where someone has to record it sitting still-is important both for physical activity and nutrition and may be used to promote increased physical activity.
• There is a need to find ways to analyze more data points and potentially richer data sets to answer research questions.
• Currently, closer monitoring of biological markers probably is better developed for the physical-activity area than for nutrition, although real-time glucose monitoring that can be linked to food intake is used in chronic diseases such as diabetes.
• Expanded tailoring beyond the theoretical constructs to environmental and social contexts can be enabled by assessment and linked with behavioral data; thereby lessening the potential for measurement error, facilitating the detection of smaller effects, and facilitating disciplinary approaches.
• Linking to GPS-type data sources can facilitate community mapping; locating individuals as they move around; knowing where they are, what they are doing, and how it relates to targeted healthy recommendations (subject, of course, to privacy concerns).

Barriers and challenges identified by the Intervention Group included:

• Providing the evidence base; testing the efficacy of new ideas before they are disseminated
• Privacy issues, especially with regard to passive measurement technologies
• Characteristics of people who may or may not accept interventions and the need to reach beyond those who are motivated to impact health problems
• Preventing technology from "running away" with or driving the field
• Lack of research, money, and time
• Need to design assessments to identify what went wrong
• Unrealistic expectations
• Unknown technology risks
• The need to learn to work with different partners such as engineers, computer programmers, marketers, etc. for implementation and funding
• The reality of field conditions such as loss, breakage, getting devices returned (particularly when working with disadvantaged groups)
• The need to target methods for reaching the underserved, the young, and the elderly
• The need to clearly think through what is to be accomplished and the amount of information to be given and received to ensure simplicity and accuracy
• Training of research staff and participants; delivery of information
• The challenge of combining with and strengthening current interventions (e.g., as enhancements/reinforcements for multiple counseling sessions or group activities)
• The need to integrate diet and physical activity.

Dr. Glanz noted that the strategies and resources needed to further incorporate real-time methodology and related technology in physical activity and/or diet research discussed by the group include:

• Formative or lead-up studies are needed, along with funding sources such as special Requests for Applications (RFAs) or administrative small grants that do not have long waiting times, and that would allow researchers to write more competitive grants, especially for interventions.
• Peers need to be exposed to what is available.
• Developments must be communicated.
• People in different fields and sectors should get together.
• Dissemination should be flexible and adaptable (e.g., templates could be created).

In addition, consortia or collaborations for both assessment and intervention could be developed, not necessarily in the sense of having a large RFA to fund everyone, but in terms of pooling resources to draw from. Examples include the NIH and the National Science Foundation (NSF) providing assistance to qualified investigators; the NIH partnering with technology companies so that, for example, researchers could obtain free equipment for small pilot studies; institutional partnerships among funders; and encouraging open code. The group suggested thinking differently from the traditional research model in terms of encouraging people to share information along the way; and to be more collaborative and less competitive in a way that is supported institutionally.

Dr. Ainsworth explained that the Assessment Group noted the dichotomy between tools that are available for researchers and tools that can assess these patterns for practitioners. The group felt that unobtrusive, easy-to-use, wearable sensors are valuable ways to measure physical activity. These might include a motion detector, such as an accelerometer or pedometer integrated with a heart-rate monitor, that measures both movement and physiologic response. Electronic diaries were mentioned, but such devices place a burden on respondents. Cameras also were mentioned as a way to measure and provide a picture of food. Cellular telephone technology with built-in sensors also was discussed. Cellular telephones already are accepted by the population. They can be adapted to capture pictures of food, and to include accelerometers and similar devices that could be downloaded easily into a database. Physiological precursor measures also were mentioned, such as a sensor on the skin or an implantable sensor that could detect different molecular structures and track changes in glucose, insulin, or the like after eating to monitor patterns. The group agreed that measures need to be unobtrusive, easy to use, and cause little burden on respondents.

With regard to uses and challenges of real-time technology, the group noted that the main types of difficulties that can be avoided by real-time assessments are respondent burden and memory difficulties common to self-report measures. Barriers to use include funding and instrument validity. The instruments are expensive, and it may be difficult to obtain adequate resources and support. Studies should be funded to validate the instruments in different populations. Perhaps a repository of instruments can be developed that researchers could borrow for use in studies. A U.S. Centers for Disease Control and Prevention physical activity group has purchased a number of accelerometers and other instruments that may be loaned to people in different countries and sites to use in studies. This has been a tremendous resource. Another suggestion was to develop a technology and assessment center for diet and nutrition from which researchers could borrow equipment and obtain answers to questions and suggestions for resources. The feasibility of using many of these instruments is a deterrent because the instruments are complicated and produce a tremendous amount of data that must be downloaded and analyzed. Additional challenges include the fact that traditional funding sources may be somewhat reluctant to fund new, innovative ideas. Dissemination strategies are needed as well, so that once an assessment method has been proven it can translated for use by practitioners in the field.

Dr. Ainsworth noted that new strategies needed to incorporate real-time methodology include:

• Dissemination strategies (see above)
• A willingness to commit resources to test, validate, and determine the acceptability of these measures
• A depository for rental or use of assessment instruments
• Support for pilot studies to develop the technology
• The establishment of a database to link food bar codes and nutrients
• The development of instruments that can assess compliance with physical activity guidelines on a large scale.

Following Small Working Group Session 1, the Intervention Working Group was split into a Physical Activity Intervention Working Group and a Dietary Intervention Working Group. Similarly, the Assessment Working Group was split into a Physical Assessment Working Group and a Dietary Assessment Working Group. All four subgroups were asked to respond to the following questions during the second small group working session:

• What are the research issues unique to physical activity (or diet) assessment (or intervention) and real-time methods/technology?



• How can these unique issues be addressed within the context of physical activity (or diet) assessment (or intervention) and real-time methods/technology?



• What recommendation/future direction is most pertinent to your specific area (i.e., diet assessment, diet intervention, physical-activity assessment, physical-activity intervention) with respect to real-time research?

Dr. James Sallis, Professor of Psychology at San Diego State University, summarized discussions from the Physical Activity Intervention Working Group. With regard to the first question on research issues that are unique to physical-activity intervention using real-time methods/technology, one of the things the group thought was unique about activity in comparison to diet was the opportunity for objective measures that do not take any effort on the part of the person. The group compiled examples of how to integrate accelerometer or pedometer data into an intervention by giving people automatic incentives and feedback throughout the day. For example, the feedback might tell someone that he or she has not been very active yet that day and that he or she should plan to do something in the future. Another example is that when people plan activity and note that in their PDAs, they can be prompted shortly before their planned session and asked if they still plan to do the activity. Tailored feedback based on the answer then can be given, and they can be rewarded afterward.

The group also established a simple framework of thinking about people, time, and place and considering how activity might have specific issues in each. In terms of time, physical activity might involve intermittent issues or patterns in which people may be active some of the time and inactive some of the time. This is similar to diet and would need to be addressed. With regard to sedentary behavior, all 24 hours of the day and how sleep is integrated into that as well as what is more typically thought of as sedentary behavior should be considered. The need to deal with all 24 hours is somewhat unique, and there are unique aspects of place for physical activity that relate to intervention opportunities. For example, kiosks could be placed in an area to inform people about the opportunities for physical activity and those would be different than the opportunities for eating. Placing sensors on recreational facilities or kiosks to orient people to places where they can be active would involve working with different partners than in the field of nutrition. Instead of working with restaurants, for example, one would work with park and recreation professionals. Intensity of activity is another consideration. There may be an issue of activity having to be of a certain intensity to derive specific health benefits. The purposes of activity, although different from nutrition, include for fun or leisure, for occupation, for transportation, for household work, and more. These different contexts for activity create different opportunities for intervention.

In terms of driving an integrated approach to activity and nutrition, the group considered the energy balance approach to be promising as an organizing tool. For example, if an individual on a given day has been eating a lot and was not very active in the morning, he or she can be given feedback indicating the amount of activity needed to keep their activity and nutrition levels in balance. With regard to place, the group thought that there was a need to provide real-time interventions both for individuals and for the environment. With individuals, PDAs or cellular telephones could be used to store information and provide feedback. With the environment, kiosks might be located at subway-station exits to provide information about nearby eating and activity opportunities. Once a restaurant is selected, information could be provided on healthy eating choices, other restaurants that might involve a longer walk and more calories burned, and so on.

The group agreed on the importance of ensuring that the technologies and methods used are applicable to the people most in need, who typically are the disadvantaged. This might be accomplished by starting with technologies that already are familiar to these populations (e.g., telephone). Additionally, the group noted the need to develop ideas for cross-training diet and physical-activity professionals in using real-time interventions. Funding studies that focus on joint development of activity and nutrition interventions would help to ensure the integration of these approaches.

Dr. Linda Nebeling, Chief of the Health Promotion Research Branch, Behavioral Research Program, DCCPS, NCI, reported on the work of the Dietary Intervention Working Group. In terms of research issues that are unique to dietary intervention, the group discussed the complexity of the components involved in dietary assessment, and the complications involved in determining appropriate assessment parameters (e.g., portion size, taste buds, jaw motion) and food components. The fact that the goal is to have people refrain from or do less of an activity rather than doing more also influences how interventions are conceptualized (e.g., a goal would be to avoid being viewed as the "food police").

The group agreed that there are many possible applications for technology. Computer technology was cited as an example of a way to engage people and keep them involved with the assessment itself in the interventions. Considerations include making applications user friendly and entertaining, particularly for population groups such as children or the elderly; the level of goal setting; decision process; developing cues and incentives most appropriate for the type of intervention designed; the potential of subject burden; food preparation; portion size; the types of brands; monitoring; tracking; dealing with different venues of information in the data stream; and how the technology itself can simplify issues of adherence and help educate users about choices. Also of concern were issues surrounding interventions and strategies and the way that technology might influence knowledge and knowledge might influence behavior at a delivery versus behavior-change level. There also are issues concerning food technology. Developments in the larger commercial environment may influence food taste, product appeal, branding, and the like; how might technology not only keep up with these, but perhaps help promote traditionally neglected food groups such as fruits and vegetables?

The need for new paradigms and exploring issues beyond the existing parameters (e.g., the concept that people eat for pleasure and the issue of society being salt-, fat-, and sugar-focused) also were discussed. Issues of monitoring or intervention were cited (e.g., in tracking behavior change, how can one avoid or account for influencing the behavior itself?). The group also expressed interest in using technology to track issues regarding satiety and hunger senses; self-ratings for intake by fullness or satiety to help generate a more mindful state of eating; and ways to account for cultural issues, including social, racial, and parental issues.

Unique issues to be addressed within the context of dietary intervention and real-time methods/ technology that the group identified include:

• Complexity in developing complete databases
• Developing more effective partnerships between the technical and application areas so that those who are experienced in working in the interventions may provide better input into how the technology is framed and developed
• Framing data that are reported (e.g., by eating occasions, from memory, by reconstruction, by event time, etc.)
• Accounting for mood states, emotional eating, and the like, as well as various related challenges.

With regard to recommendations and future directions, the group made the following recommendations:

• Developing cross-technology centers that would combine public-health applications with areas funded by the NSF
• Enhancing the areas of reinforcement and incentives (which relates to physical activity and nutrition), including ways to generate rewards (e.g., via Web TV), and impulse purchases and sensor warnings
• Funding needs and ways to use technology to streamline areas so that there might be three to five regional centers that would help develop and facilitate partnerships in such areas as food purchase and food procurement and collaborations with the NIH regarding nutrition and behavior training and research.

Dr. Ainsworth explained that the group discussed determining the unit of measure for physical activity and whether it should be energy expenditure or physical activity and movement at light, moderate, or vigorous intensities. Another topic was the difficulty of measuring absolute energy expenditure. There is a need for methods to refine the estimation of energy expenditure from physical activity in real time, and real-time units are needed to indicate intensity of motion. The most likely scenario may be a combined motion sensor and heart-rate monitor that is capable of measuring both movement and physiologic response. As they are developed, these new sensors should be tested in multiple populations and locations so that sufficient databases can be developed to enable practitioners to use these tools to monitor their own behavior.

Dr. Arab served as the spokesperson for the Dietary Assessment Working Group. She indicated that the group identified seven areas that distinguish dietary assessment from physical activity in terms of the needs in this area:

• Infrequent consumption refers to foods that may be eaten infrequently but still are important to capture because they may be particularly healthful or perhaps toxic. Although infrequent physical activity events may be irrelevant, that is not the case with dietary consumption. Thus, many data points that are interspersed over longer periods of time are needed for dietary assessment.
• Food identification refers to the difficulties researchers have in describing what they want to capture and individuals have in describing what they eat. Various foods may be difficult to distinguish when looking at a picture.
• Portion sizes are difficult to identify and describe.
• Nutrient databases are difficult to establish because diets can be extremely complex.
• Dietary patterning includes the need to know not just if a particular food was eaten but in what context it was eaten and what else might have been eaten with it.
• Special behaviors refers to such actions as food binging and going to the refrigerator in the middle of the night. These are behaviors that individuals may not want researchers to know about.
• Seasonality is another issue that is somewhat unique to dietary assessment.

With regard to whether technology could address any of the above, the group noted the following:

• Addressing infrequent consumption would involve getting and keeping the camera rolling, having the camera be unobtrusive and affordable, and managing the data collected.
• The group did not feel technology would be too helpful in addressing food identification issues. For portion sizes, however, new technologies can be tremendously helpful in allowing much greater precision.
• For the nutrient database problems, it was unclear whether technology could help. Perhaps if biomarkers of what is being consumed are involved, technology could be helpful.
• Dietary patterns can be captured more readily with new technologies because of the various times of the day that food is consumed, there is the potential for more contextual data surrounding the vision of what foods are eaten together, and observations can be conducted over longer periods of time.
• Special behaviors may be better addressed via cameras.
• It may be possible to address seasonality more thoroughly via advances in technology.

In terms of recommendations or future directions that are most pertinent to this area, the group felt that the "specific area" could vary, for example from total diet to fruits and vegetables to heterocyclic amines. Different methods may be required for different questions. The group agreed that technology should be easy and relatively unobtrusive. Problems cited by the group included the possibility of losing devices and what happens to the data on lost devices. Privacy protection was another issue raised by the group. The group also discussed the issue that many dietary interest areas reflect long-term diet or diet at specific times of life, and that the technologies mentioned do not capture retrospective assessment, which is a weakness. The need to approach the task of dietary assessment from multiple perspectives was discussed, perhaps by using pictures, verbiage, and followup questions to obtain a complete picture instead of relying on a single approach.

Dr. Rachel Ballard-Barbash, Associate Director, Applied Research Program, DCCPS, NCI, also noted that the group discussed the potential for technology to help in deconstructing the cognitive tasks related to recall. At present, cognitive interviewing primarily is conducted in a laboratory where people are asked questions. Perhaps there is some potential for using this type of technology within the construct of self-report (i.e., to ask people questions as they are reporting when problems or issues arise). The group agreed that technology might help improve existing measures that are used in very large population samples. Dr. Arab added that the group also discussed using these methods to validate and calibrate less expensive and demanding methods, perhaps not as a replacement, but as a support.

Dr. Atienza asked Dr. Scott Leischow (Acting Associate Director, Behavioral Research Program, DCCPS, NCI); Dr. Gary Kreps (Chief, Health Communication and Informatics Research Branch, Behavioral Research Program, DCCPS, NCI); and Fran Thompson (Nutritional Epidemiologist, Applied Research Program, DCCPS, NCI) to assemble as an NCI panel to react to the ideas, recommendations, and challenges that were discussed during the working group meetings and presented by group spokespersons.

Dr. Kreps noted that there were a number of issues discussed that cut across both assessment and intervention, and that cut across both dietary issues and exercise and physical activity. He discussed general needs, issues, opportunities, and some challenges related to these crosscutting issues. Dr. Kreps explained that there is a tremendous need for good descriptive data about both dietary behavior and exercise behavior. He noted that there are significant limitations in the ways that these behaviors have been studied in the past, and there are gaps in knowledge. It is difficult to develop effective, powerful, and meaningful interventions when what people are doing, why, and how. This creates opportunities with new technologies and new ways of gathering data to provide the baseline database that will allow for a more complete understanding. Furthermore, he noted a tremendous need for powerful and appropriate interventions that speak to individuals and where they are; this need cannot be addressed until further knowledge is gained. Ways to energize understanding of these phenomena and possible interventions are needed.

Dr. Kreps indicated that real-time data collection and the use of new information technologies afford concrete opportunities. These include real-time point-of-purchase information as it happens; reduced self-response bias; nonreactive and unobtrusive data; real-time monitoring and self-monitoring of behavioral activities; and opportunities to gather and provide interactive information, probe for additional data, and obtain different levels of information about dietary and exercise behavior. It also facilitates testing the strengths and difficulties of different methods, and enables clarification of the kinds of data available via comparison and a multimethodological approach. He noted that it may be better to consider real-time data collection as a supplemental rather than a replacement method; it is a way of adding to the different information and data-gathering tools that are available.

Dr. Kreps commented that cost will be a significant obstacle, especially in preparing to build the infrastructure for employing these new approaches and in implementing and adapting the technologies to meet the unique needs of this area. Researchers are behind the "technology curve" in terms of the research being done, and the technologies were not necessarily developed with researchers in mind, so there is a need to both adopt and adapt the technologies to meet researchers' data-gathering needs. Analyzing the extensive data that are gathered (e.g., from 24 hours per day/7 days per week monitoring of health behavior in large groups of people) is another challenge. Researchers will need to develop new analytic strategies, new ways of coding, and new ways of providing descriptive and inferential data about findings.

There also are formidable ethical issues, particularly in terms of surveillance and intrusion and privacy, and whether the data that are collected and instruments used will be confidential. How pervasive this research is allowed to become is a major ethical consideration. These technologies will afford a major challenge and opportunity to increase the level of analysis in research being conducted. Where health-promotion researchers typically have focused on the individual level, these technologies will allow broadening that focus to larger social networks and societal and cultural influences.

Dr. Leischow noted that one challenge will be to harness and prioritize the ideas presented at this meeting within the appropriate scientific disciplines and scientific social networks. The government will play some role in this, but the task mostly will fall to the researchers involved. The challenge will be to go beyond proximal linkages to incorporate work that has occurred in fields beyond health (e.g., work done by the National Aeronautics and Space Administration, the military, business, physics, etc.).

Another challenge will be integrating the technology focus and data collection process via technology with larger national goals such Healthy People 2010. How might this work facilitate reaching such goals in a more efficient way? This may provide a justification for pursuing these techniques. Dr. Leischow noted, however, that there will be significant funding challenges, especially in view of the current leveling-off of and potential decreases in NIH funding levels and the fact that behavioral research is costly.

Dr. Leischow noted the need to think about the proposed research in terms of the three-dimensional paradigm of Discovery, Development, and Delivery; a basic systems approach that has been used in fields other than public health. Determining how surveillance might be used to optimize the movement of discoveries through the development and delivery phases and then feeding back from delivery to more new discoveries will be an interesting and important challenge.

Dr. Leischow noted that another challenge will be the relationship between corporate and public interests, especially as it pertains to technologies. Corporate interests or lack thereof may play a role in the development of new behavioral interventions. Equipment and new data collection technologies will be driven in some fashion by corporate interests, and researchers need to consider creating both firewalls and linkages between data collected by corporations and data collected by scientists.

Dr. Thompson emphasized the importance of clarifying objectives. These technologies can be applied toward understanding behaviors, in assessing diet or physical activity, and also in intervention. All of these applications are valuable, and researchers need to be clear about their own objectives and what they want to accomplish. She also highlighted the importance of the energy-balance concept and noted that, although diet and physical activity differ and diet is considered to be harder to measure than physical activity, diet and physical activity share many characteristics. It would be beneficial for researchers to work together on issues of physical activity and diet, and to form interdisciplinary groups to carry out such activities. There is a large gap between the current state-of-the-art and where researchers would like to be in the future. It would be useful to identify small steps that can be taken now to reach some ultimate and defined objective. Identifying those steps will facilitate determining how to get there and smaller studies of the validity of methods used on the way.

Dr. Glanz asked whether it is possible to waterproof sensors to measure waterborne activity. Dr. Ainsworth noted that there are waterproof sensors. Dr. Richard Troiano, Director and Research Scientist in the Risk Factor Monitoring and Methods Branch, DCCPS, NCI, indicated that activities such as swimming may require multiple sensors (i.e., such as one on the arm and one on the leg). Dr. Intille added that for showering and similar activities, a waterproof transmitter in combination with a receiver such as a cellular telephone or a PDA placed somewhere in the bathroom would be sufficient to record activity.

Dr. Atienza asked Drs. Ballard-Barbash and Nebeling to discuss their reactions to the ideas and recommendations that surfaced concerning energy balance. Dr. Ballard-Barbash indicated her belief that it would be useful for researchers in the areas of dietary and physical-activity assessment to work together in various areas, including that of measurement error. It may be in part because measurement error is such an enormous problem in diet that people have put more energy into that field, but some of those issues might be useful and relevant in the area of physical activity as well. She noted her agreement with the notion that this is not a replacement technology, but a technology with many utilities that must be applied in the most appropriate settings. She also indicated that there may be enormous utility in applying these techniques to subsamples within larger studies when it does not seem appropriate to apply the technologies to a very large sample or population, perhaps as a way to validate or calibrate or look at some of the other simpler measures that may be used. The techniques may be used in this way for surveillance and perhaps for large epidemiologic studies. Dr. Ballard-Barbash indicated that one of the greatest areas of promise may be the field of intervention because providing feedback on a real-time basis may promote progress in behavior change. Integrating feedback on diet and physical activity is an exciting prospect that may help people to better manage their weight. Applying these techniques in environmental ways, particularly in terms of creating ways of monitoring community-level behavior that can be fed back more rapidly, is another area of interest. Dr. Nebeling agreed, noting that her group hopes to enhance the capacity to integrate these discipline areas to improve the validation of this technology and its application in the intervention environment to facilitate behavior change, the overall objective.

Dr. William Haskell, Professor of Medicine at Stanford University, stated that he is aware of no systematic trials in the area of physical activity that have been conducted to show that EMA added to a physical activity intervention increases the change in physical activity. He asked whether anyone else knows of such a study. Although there is an EMA pedometer, there do not appear to be any studies of physical-activity intervention in which one-half of the sample was randomized to use the pedometer, and the other half were withheld the pedometer to see whether it made a difference. Dr. Glanz noted that in the nutrition area, there was a small, nonrandomized study that involved immediate feedback, weekly feedback by letter, and cumulative feedback, and the women did come closer to reaching their goals and were shown to have reduced their fat and calorie intake. They did not change their fruit and vegetable intake, however

Dr. Troiano raised the issues of extending interventions to the population level because that is the level of the obesity problem, and of developing devices with a commercial market to get the devices made and produced less expensively. This involves creating a health objective based on an objective measure. He noted that currently, the field of physical activity uses vague terms that may be unclear to many people (e.g., "moderate intensity," "accumulating"). He suggested that a clear health objective, such as "spend at least a certain amount of time in a particular heart rate zone," might create market for a heart-rate monitor that can accumulate and indicate when someone is in that zone, making it easier for people to know what is meant and how they are progressing with respect to such an objective. Dr. Ainsworth added that Japan, as one of its health objectives, has a number of steps per day based on the pedometers, which may be an objective that could be implemented sooner because of the great interest in pedometers.

Dr. Sallis indicated that he did not know why there have been no studies to determine the effects of pedometer monitoring. Dr. Haskell indicated that it would be straightforward, and Dr. Ballard-Barbash stated that the study might not require a large sample size. Dr. Ainsworth said that she would check the literature. Dr. Sallis commented that real-time assessment probably has some promise in approaching energy balance, but he is not sure that it does from an intervention point of view. This is because total energy intake seems to be one of the most difficult things in diet to assess and there is uncertainty about whether the real-time method will reduce that burden. The EMA might, however, for intervention purposes be better suited for more specific dietary targets such as fruits and vegetables, calcium-rich foods, and the like. Such assessments would be simple and quick enough to be monitored several times per day over an extended period without creating a major burden.

Dr. Arab noted that researchers sometimes forget that what is being looked for is the rate of calorie intake per unit of time. What is the rate or the time that is desired and how can it be sampled? The use of modern technology does not automatically mean reaching new levels of expense. Some of the technology involves cellular telephones that already exist and are in use. There are expenses involved in handling large amounts of information, but these are not necessarily long-term hardware-related expenses and may not be a significant factor. She expressed concern that, in the dietary area, information often is sold too cheaply. What is spent on other environmental measures vastly exceeds that for a single component of health.

Dr. Patterson commented that when researchers think about diet, they sometimes set themselves up to fail by setting expectations that are too high. Considering what can be done easily and simply and how the data can be used effectively might be a better approach. For example, if a bar code could be developed, someone could wand their whole kitchen in just a few minutes and determine the caloric intake there. Household surveys have shown that inventorying what is in a person's house can be related to average intake over several days for individuals within the house. It may not be perfect, but it can be helpful. Developing such behavioral markers rather than trying to be so quantitative in thinking about diet may be a better approach.

Dr. Glanz described a disconnect on the dietary side between the assessment and intervention issues, because interventions will not be aimed at the complexity of diet, but instead at something more moderate and known-that the majority of calories come in a definable number of foods. Although there will be error, these types of measures can be done as an alternative to measuring total diet when working on the energy-balance issue. She noted that this is an important opportunity. An effort is underway to find funding to use PDAs to validate and go beyond diet assessments to address questions such as where people get their food. This type of information has not been available on a daily basis or from food-frequency questionnaires over time.

Dr. Haskell raised the issue of the value or lack of value of monitoring body weight, the outcome of interest, in terms of energy balance, perhaps done in innovative ways that may provide positive feedback. In other words, if weight could be monitored on a daily basis-although some weight-loss experts would not be in favor of that-with feedback relating what was done with change in weight over time, it would be interesting. Dr. Glanz commented that this could be the subject of an additional meeting.

With regard to costs, Dr. Haskell commented that his group had used a self-assessment cardiovascular-risk questionnaire in scannable form that cost approximately $10 per questionnaire to mail out, complete, scan, and send back at the company level. In a current Internet-based project involving some 20,000 employees, the price has dropped to about 20 cents per questionnaire. Thus, new technology does not necessarily lead to higher cost.

Dr. Intille mentioned that the trend at supermarkets is toward having individuals scan their own items. This may provide an opportunity for nutrition researchers because items will be scanned at the point of purchase. Dr. Arab commented that she thought the point of real-time scanning was to scan at the time of consumption versus time of purchase. Dr. Glanz noted that it could be either, and Dr. Intille expressed the view that the point-of-decision for food is when it is purchased. Dr. Sallis noted that this may present an intervention opportunity in that nutrient information could be incorporated into the process. Dr. Ballard-Barbash noted that most food producers perceive that as proprietal data, but that summaries could be used. Dr. Patterson commented that fast food and other restaurants could be urged to have scanners on their menus so that people could scan as they ate at restaurants. Dr. Leischow noted that companies already use similar technology when they collect information and provide coupons tailored to individual customers at checkout.

Dr. Ainsworth returned to Dr. Intille's earlier comment regarding manufacturers being willing to respond to researchers' requests with regard to developing equipment. Dr. Intille stated that manufacturers of mobile computing devices want ideas that will help them to differentiate their products. If they could say, "Our product will help you eat better," and it does so in a way that is pleasing to consumers, then they will be open to listening to such ideas. It is likely that some manufacturers will put accelerometers, for example, in their cellular telephones for gaming applications. This could be beneficial for measuring physical activity. Unless they are told by this community that that this is an area of interest, however, they may focus only on the gaming aspect. That is why it would be interesting to have some rapid prototyping via small pilot studies from this community to provide input to manufacturers during this development process.

Dr. Glanz noted that this discussion underscores the cyclical nature of the real-time assessment and intervention issue because intervention groups feel strongly that there need to be tests of the efficacy of these kinds of instruments. Although researchers may succeed in getting manufacturers to make devices, the manufacturers are not necessarily interested in whether such devices actually improve health-related behavior. There is a need to enable some studies that will provide efficacy information. Dr. Leischow asked whether some of these technologies might already have been implemented in countries such as Iceland, where there is a high degree of tracking in other domains. Iceland has genealogy records that go back many years, and there are samples on many people in the country, information on individuals' health care, and confidentiality rules that are far more stringent than in the United States. He suggested Iceland as an interesting laboratory for testing technologies in a way that allows potential exploration of interesting approaches while protecting confidentiality. Dr. Glanz noted that Luxemburg is another country that does this, and it may be more accessible. Dr. Ballard-Barbash raised the possibility that the NIH might engage with the technologic community in this area, as it has with drug companies, to promote linkages between such companies and research happening within the NIH.

Before adjourning the meeting, Dr. Atienza thanked participants for attending the meeting and complimented the group for the ideas that had been generated. He explained that he will establish a listserv for individuals who want to continue with these discussions and connect with publications, other references, other ideas, or network. A writing group comprised of meeting attendees will collate ideas discussed at the meeting, and all attendees will be given an opportunity to comment on a draft of a manuscript that will include both physical activity and dietary intervention and assessment.