This is an open-access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/2.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work, first published in the Journal of Medical Internet Research, is properly cited. The complete bibliographic information, a link to the original publication on http://www.jmir.org/, as well as this copyright and license information must be included.
The development of a mobile telephone food record has the potential to ameliorate much of the burden associated with current methods of dietary assessment. When using the mobile telephone food record, respondents capture an image of their foods and beverages before and after eating. Methods of image analysis and volume estimation allow for automatic identification and volume estimation of foods. To obtain a suitable image, all foods and beverages and a fiducial marker must be included in the image.
To evaluate a defined set of skills among adolescents and adults when using the mobile telephone food record to capture images and to compare the perceptions and preferences between adults and adolescents regarding their use of the mobile telephone food record.
We recruited 135 volunteers (78 adolescents, 57 adults) to use the mobile telephone food record for one or two meals under controlled conditions. Volunteers received instruction for using the mobile telephone food record prior to their first meal, captured images of foods and beverages before and after eating, and participated in a feedback session. We used chi-square for comparisons of the set of skills, preferences, and perceptions between the adults and adolescents, and McNemar test for comparisons within the adolescents and adults.
Adults were more likely than adolescents to include all foods and beverages in the before and after images, but both age groups had difficulty including the entire fiducial marker. Compared with adolescents, significantly more adults had to capture more than one image before (38% vs 58%,
A majority of both age groups were able to follow the defined set of skills; however, adults were less efficient when using the mobile telephone food record. Additional interactive training will likely be necessary for all users to provide extra practice in capturing images before entering a free-living situation. These results will inform age-specific development of the mobile telephone food record that may translate to a more accurate method of dietary assessment.
Dietary intake is an important environmental exposure to consider when evaluating an individual’s or population’s risk for chronic disease. A link between diet and the development of certain cancers, cardiovascular disease, liver disease, and type 2 diabetes has been established. However, scientific evidence linking diet and genetics to these diseases continues to emerge [
Dietary assessment is difficult due to the increasing complexity of the food supply and day-to-day variability in a person’s diet [
Researchers have been striving to harness the potential of new digital technologies to improve the effectiveness of their work, and researchers in the field of dietary assessment are no different. The past 10 to 15 years has seen steadily increasing usage of mobile communication devices [
However, with the rapid advancement in the capabilities of mobile devices, researchers are now pursuing image-based methods as a way of addressing the limitations of traditional dietary assessment methods [
The design of the mobile telephone food record has been described previously [
There are challenges related to using smart phones in this new dietary assessment method. For example, for adolescents to use the device, school administrators must accept its use on the school campus, as young people are in school most days of the week. Adults are often less facile than adolescents with using new technology. Therefore, the mobile telephone food record design needs to address these concerns.
Evidence-based development is a crucial step in designing the mobile telephone food record for use by both adults and adolescents [
The objectives of this study were to evaluate a defined set of user skills for both adults and adolescents—that is, successful image capturing of an eating occasion, while using the mobile telephone food record—and to compare the perceptions and preferences between adults and adolescents regarding their use of the mobile telephone food record. A priori, our hypothesis was that statistically significant differences between adults and adolescents would emerge that would need to be translated into different mobile telephone food record designs to accommodate lifestyles and abilities to use a new technology.
We collected data from two samples of adolescent participants [
The first adolescent sample was drawn from summer camps for adolescents, ages 11–18 years, taking place on the campus of Purdue University in 2008. A total of 63 participants from these camps used the mobile telephone food record for meal session 1, and 55 (87%) returned for meal session 2 the following day. After using the mobile telephone food record for meal session 1, participants provided feedback and received additional training during the postmeal 1 session. During this session, the participants responded to a series of statements regarding their perceptions of the mobile telephone food record and preferences when using the mobile telephone food record. The advanced interactive instruction included activities in which the participants practiced taking images in potentially problematic snacking scenarios.
The second adolescent sample was a convenience sample drawn from the local community [
The adult sample was a convenience sample drawn from the campus of Purdue University and the local community during the fall of 2008. A total of 57 participants, ages 21–65 years, used the mobile telephone food record for meal session 1, and 24 (42%) returned for meal session 2 on a subsequent day (
Study design, activities, and measures of participants using the mobile telephone food record. For 15 of the adolescent participants, meal session 2 was later in the same day. For the remainder of participants, meal session 2 occurred on a different day. Adult participants were offered dessert as a separate course. For meal session 1, 39 selected dessert, and for meal session 2, 15 selected dessert.
The menus served to the adolescents have been described previously [
We used HTC p4351 mobile telephones (HTC Corporation, Taoyuan, Taiwan) running Windows Mobile 6.0 Professional (2007; Microsoft Corporation, Redmond, WA, USA). The software, described previously [
Images that demonstrate meeting two skills required for using the mobile telephone food record: included in the image are all foods and beverages and the entire fiducial marker (checkerboard square).
To assess the two skills of including all foods and beverages and the entire fiducial marker in the image, the before and after meal images were evaluated for the inclusion of these two required items. When evaluating the inclusion of all foods and beverages, the images were coded as yes if all of the foods and beverages were visible in the image, no if any of the food or beverage was not visible, or software programming error if the image was unavailable due to software malfunction. When evaluating for the inclusion of the entire fiducial marker, the images were coded as yes if the entire fiducial marker was visible in the image, no if a portion of the fiducial marker was cut off, or software programming error. To evaluate the skill of efficiently taking only one image, the number of images taken by each participant before and after meal sessions was coded as one image or greater than one image.
During the feedback sessions, we showed statements regarding possible perceptions of the mobile telephone food record and preferences when using the mobile telephone food record using PowerPoint (Office 2007, PowerPoint 2007; Microsoft Corporation). The participants responded to these statements using a 5-category ordinal response scale (ie, strongly agree, agree, neutral, disagree, and strongly disagree).
We showed the following five statements regarding perceptions to all participants: (1) I think it would be easy to remember to take an image before meals, (2) I think it would be easy to remember to take an image after meals, (3) I think it would be easy to remember to take an image before snacks, (4) I think it would be easy to remember to take an image after snacks, and (5) the software was easy to use.
We showed the following four statements regarding preferences to all participants: (1) I think it would be easy to carry and use a credit card-sized fiducial marker, (2) I think it would be easy to carry and use a USB-sized fiducial marker (to denote size, this was defined to participants as USB flash drive, USB memory stick, USB jump drive, or USB thumb drive), (3) I prefer to stand while taking an image, and (4) I prefer to sit while taking an image.
The adolescents responded to these nine statements at the start of the postmeal 1 session, followed by the advanced interactive instruction. The adolescents’ responses were collected with the
During the premeal session, participants in the adult sample responded to perception statements 1–4. We asked the adults questions to assess their previous experience capturing images with digital cameras and mobile telephones. These were (1) Do you own a digital camera? (2) How often have you taken pictures with a digital camera? and (3) How often have you taken pictures with a mobile telephone?
The response choices for these latter questions were frequently, occasionally, and never or rarely. The adult participants responded to the nine statements above in the postmeal 1 session. The adults recorded their responses on a paper form.
We used data that we collected using the same methods among the adults and the adolescents for statistical comparisons. To further delineate differences by age, we divided the adolescent sample into early and late adolescence: 11–14 years and 15–18 years, respectively. The adult sample was divided into early and middle adulthood: 21–40 years and 41–65 years, respectively. Descriptive analysis included frequencies and percentages. Within both the adolescent and the adult samples, we analyzed differences in age groups and gender using chi-square. McNemar test was used for comparisons of the set of skills for capturing images within the adolescents and within the adults. For those comparisons, each skill (eg, all foods being in image) was classified as yes (demonstrating the skill) or no (not demonstrating the skill). Chi-square was used for comparisons of the skill set between the adults and adolescents; for these comparisons, we grouped no and software programming error together. The 5-category ordinal response scales used by the participants to provide their preferences and perceptions were recoded as agree, neutral, or disagree. We compared perceptions and preferences between adults and adolescents using chi-square. For comparisons with an expected cell count of less than 5, limiting the comparison to agree and disagree eliminated the inadequate cell counts. We used SPSS 17.0 (IBM Corporation, Somers, NY, USA) for all statistical analyses.
A total of 135 participants (78 adolescents, 57 adults) used the mobile telephone food record for meal session 1, and 94 (70 adolescents, 24 adults) returned to use the mobile telephone food record for meal session 2. The descriptive characteristics of both samples are in
Software programming errors occurred when saving the image on the mobile telephone food record, making them unavailable for the analysis. These errors resulted in partial loss of images, either a before or an after image; however, no images were available for only one adult participant, leaving 56 adults for this analysis. Changes to the software were made after testing it with the adolescents, which likely accounted for the reduction in programming errors experienced by the adults.
For both adults and adolescents, inclusion of the fiducial marker in the image was more problematic than inclusion of all of the foods and beverages (
Adolescents had a stronger preference than adults for the size of the fiducial marker that they would be willing to use (
Characteristics of adults and adolescents testing the usability of the mobile telephone food record.
Characteristic | Adolescents |
Adults |
|
|
|||
Male | 26 (33%) | 18 (32%) | |
Female | 52 (67%) | 39 (68%) | |
|
|||
11–14 | 45 (58%) | NAa | |
15–18 | 33 (42%) | NA | |
21–40 | NA | 27 (47%) | |
41–65 | NA | 30 (53%) | |
|
|||
Asian | 1 (1%) | 4 (7%) | |
Hispanic | 7 (9%) | 0 (0%) | |
Non-Hispanic white | 55 (70%) | 45 (79%) | |
Black/African American | 10 (13%) | 2 (4%) | |
Multiple | 5 (6%) | 6 (11%) |
a Not applicable.
Evaluation of participants’ set of skills when capturing images with the mobile telephone food record.
Skill | Adolescents (n = 78) | Adults (n = 56)a | |||||
Yes, |
No, |
Software |
Yes |
No, |
Software |
||
|
|||||||
Meal session 1c | 61 (78%) | 7 (9%) | 10 (13%) | 53 (95%) | 0 (0%) | 3 (5%) | |
Meal session 2 | 59 (84%) | 9 (13%) | 2 (3%) | 23 (96%) | 0 (0%) | 1 (4%) | |
Dessert session 1d | NAe | NA | NA | 39 (100%) | 0 (0%) | 0 (0%) | |
Dessert session 2d | NA | NA | NA | 14 (93%) | 0 (0%) | 1 (7%) | |
|
|||||||
Meal session 1 | 54 (69%) | 14 (18%) | 10 (13%) | 44 (79%) | 9 (16%) | 3 (5%) | |
Meal session 2 | 53 (76%) | 15 (21%) | 2 (3%) | 18 (75%) | 5 (21%) | 1 (4%) | |
Dessert session 1d | NA | NA | NA | 37 (95%) | 2 (5%) | 0 (0%) | |
Dessert session 2d | NA | NA | NA | 11 (73%) | 3 (20%) | 1 (7%) |
a Due to software programming error, n = 56 instead of 57.
b Paired images unavailable due to software programming errors.
c
d Dessert was served as a separate course for adult participants. For meal session 1, 39 selected dessert, and for meal session 2, 15 selected dessert.
e Not applicable.
Comparisons between and within adolescents and adults of the number of images acquired prior to obtaining a suitable image.
Group | Adolescents |
Adults |
|||||||
1 image, |
>1 image, |
Data recording |
1 image, |
>1 image, |
Data recording |
||||
|
|||||||||
Meal session 1 | |||||||||
Before imagec,d | 38 (62%) | 23 (38%) | 2 | 21 (42%) | 29 (58%) | 6 | |||
After imagec,e | 44 (75%) | 15 (25%) | 4 | 25 (50%) | 25 (50%) | 6 | |||
Meal session 2 | |||||||||
Before image | 39 (77%) | 12 (24%) | 4 | 13 (59%) | 9 (41%) | 2 | |||
After image | 40 (78%) | 11 (22%) | 4 | 16 (73%) | 6 (27%) | 0 | |||
|
|||||||||
Meal session 1 | |||||||||
Before image | 28 (58%) | 20 (42%)g,h | NAi | 9 (45%) | 11 (55%) | NA | |||
After imagec,j | 36 (75%) | 12 (25%) | NA | 7 (35%) | 13 (65%) | NA | |||
Meal session 2 | |||||||||
Before image | 38 (79%) | 10 (21%)g,h | NA | 12 (60%) | 8 (40%) | NA | |||
After image | 37 (77%) | 11 (23%) | NA | 14 (70%) | 6 (30%) | NA |
a Due to software programming errors, n = 56 instead of 57.
b Data recording error on the part of staff; therefore, numbers not included in percentages, which represent only users’ abilities.
c Comparison between adolescents and adults.
d
e
f Number of before and after meal images these participants took was recorded for both meal session 1 and meal session 2 (n = 48 session pairs for adolescents; n = 20 session pairs for adults).
g Comparison between meal session 1 (before) and meal session 2 (before) within adolescents.
h
i Not applicable.
j
Comparison of perceptions and preferences between adolescents and adults regarding use of the mobile telephone food recorda.
Perceptions and preferences | Adolescents (n = 78)b | Adults (n = 57) | |||||
Agree, |
Neutral, |
Disagree, |
Agree, |
Neutral, |
Disagree, |
||
|
|||||||
The software was easy to use | 55 (71%) | 9 (13%) | 6 (9%) | 52 (91%) | 1 (2%) | 4 (7%) | |
I think it would be easy to remember to take an image before mealsc | 26 (37%) | 22 (31%) | 22 (31%) | 47 (83%) | 5 (9%) | 5 (9%) | |
I think it would be easy to remember to take an image after mealsc | 29 (41%) | 27 (38%) | 15 (21%) | 42 (74%) | 8 (14%) | 7 (12%) | |
I think it would be easy to remember to take an image before snacks | 8 (11%) | 16 (23%) | 46 (66%) | 15 (26%) | 12 (21%) | 30 (53%) | |
I think it would be easy to remember to take an image after snacks | 15 (21%) | 19 (27%) | 37 (52%) | 19 (33%) | 13 (23%) | 25 (44%) | |
|
|||||||
I think it would be easy to carry and use a credit card-sized fiducial marker | 55 (77%) | 10 (14%) | 6 (8%) | 52 (91%) | 4 (7%) | 1 (2%) | |
I think it would be easy to carry and use a USB-sized fiducial markerd | 30 (42%) | 19 (27%) | 22 (31%) | 38 (67%) | 15 (26%) | 4 (7%) | |
I prefer to stand while taking an imagec | 43 (63%) | 14 (21%) | 11 (16%) | 13 (23%) | 12 (21%) | 32 (56%) | |
I prefer to sit while taking an imaged | 25 (36%) | 21 (30%) | 23 (33%) | 39 (68%) | 8 (14%) | 10 (18%) |
a Percentages do not add to 100% due to rounding.
b Missing values due to a malfunction of the
c
d
This is the first study to systematically evaluate the abilities of adolescents and adults to provide accurate images of an eating occasion. A priori, we assumed that huge differences in skills with technology between adolescents and adults would emerge; however, other than number of images captured, nothing else became obvious. The adolescents were more efficient: they took fewer images than the adults. By the second meal, the adolescents became even more efficient, whereas the adults made insignificant gains. Also, by the second meal, the inclusion of all foods in the images was the same between adults and adolescents, whereas inclusion of the important nonedible item (ie, the fiducial marker) was more problematic for both adolescents and adults. These results support that the fiducial marker was too large. As such, it was difficult to include in images without being partially covered by a plate or utensil. Evaluation of the images for the placement of the fiducial marker revealed that the participants placed the fiducial marker in various locations in the meal setting. Thus, work to reduce the size of the fiducial marker is justified. For both age groups, a notification from the device that the entire fiducial marker is not in the camera’s field of view may be helpful in reminding participants to include the entire fiducial marker when capturing images. Clear instruction on the desired placement of the fiducial marker may prevent the participant from spending time deciding where to locate it in the meal setting, which might reduce the burden of this task and translate to better cooperation with this step.
We have also established that the perceptions and preferences of adolescents and adults regarding use of the mobile telephone food record were more disparate than their skill set. In particular, adolescents were less likely than adults to agree that capturing images of meals before and after would be easy. Adolescents were more opinionated about preferring a credit card-sized fiducial marker. The adolescents may have preferred a credit card-sized fiducial marker because it could be easily carried in a wallet. Finally, adolescents stated a preference to stand while using the mobile telephone food record and adults preferred sitting. This preference for standing is consistent with irregular eating patterns and selecting snacks that are easily portable and often eaten while standing [
Adolescents are typically the earlier and more eager adopters of new technology [
Despite the adolescents being observed as more confident and comfortable when using the mobile telephone food record, they were less likely to agree that it was easy to capture pre and post meal images. This could be a result of differences in daily schedules between the two age groups and may reflect adolescents having more irregular meal times than adults [
Based on the length of time between the before and after images, the average meal duration was shorter for adolescents than for adults. This information provides a basis for programming age-specific software to start timing after the first meal image is captured to initiate a reminder for taking the after image. Next steps include testing the mobile telephone food record with participants in a free-living environment to ascertain the true level of burden, duration of cooperation, and accuracy of recorded energy intake using a biomarker for energy, such as doubly labeled water. There were minimal differences regarding preferences, which will simplify the design process for the mobile telephone food record for adults and adolescents.
The results of these studies will translate to minimal design differences of the mobile telephone food record between adolescents and adults. The majority of both adults and adolescents were able to follow the defined set of skills when capturing before and after images of their meals; however, these results do provide evidence for the need for some age-specific development of the mobile telephone food record, such as reminder programming. The adults were more cautious than the adolescents when taking images and as a result were more likely to include all food and the fiducial marker, which are necessary to capture an image suitable for image analysis. However, adults had to take more images than adolescents before capturing satisfactory ones. Although they were less efficient, the adults perceived that remembering to capture images with the mobile telephone food record would be easy. Additional use of the mobile telephone food record improved adolescents’ perceptions and set of skills when capturing images. Additional interactive training will likely be necessary for all users to provide extra practice in taking images before entering a free-living situation. The adolescents had a stronger opinion about the size of the fiducial marker than the adults, suggesting that the fiducial marker design needs to accommodate adolescents over adults. Software improvements between the adolescent and adult meal sessions greatly reduced the number of software programming errors. Some problems will likely never be entirely eliminated due to low battery power and other software-related difficulties, but advances in technology will ensure that these errors will become less frequent.
A more accurate method of dietary assessment will help strengthen the ability of researchers to identify diet–disease and diet–gene relationships. The data generated from a tool such as the mobile telephone food record could be combined with measures of the built environment to inform public policy and assist in the development of nutrition interventions. Further, novel dietary assessment methods will contribute to the growth of mobile applications to enhance self-monitoring for diabetes, weight control, and other diet-related diseases.
mobile telephone food record
personal digital assistant
Support for this work came from the National Cancer Institute (NCI; 1U01CA130784-01), the National Institute of Diabetes and Digestive and Kidney Diseases (NIDDK; 1R01-DK073711-01A1), and the NCI, Nutritional and Behavioral Cancer Prevention in a Multiethnic Population postdoctoral fellowship (R25 CA 90956; REG). The contents of this publication do not necessarily reflect the views or policies of the NCI or NIDDK, nor does mention of trade names, commercial products, or organizations imply endorsement from the US government.
All authors contributed equally to this work.
None declared.