Designing food diaries to support collaboration between individuals and health experts
Many people keep a record of what they eat. Some people do it to manage weight; others want to monitor their health conditions; still others try to understand their everyday eating. While food diaries have the potential to support a variety of health goals, many people struggle with recording and making sense of their everyday food intake. Facing these challenges, people increasingly turn to health experts — doctors, dietitians, nutritionists, health coaches — to help analyze the data and provide recommendations.
In our previous research, we found that this sort of collaborative review of personal data can help people, but it is hard. Designs that help people and health experts to communicate individual goals, include individual preferences and constraints into conversations, and that effectively summarize tracked data best facilitate helpful conversations, leading to individualized interpretations and actionable recommendations.
Through our design and evaluation of a photo-based food diary — Foodprint — we demonstrate how a photo-based diary and visual summary system can be designed to help people and health experts individually and collaboratively reflect on everyday food data, identify healthy eating strategies, and develop symptom management plans. We also show how these designs can help individuals and health experts explicitly include individual goals and knowledge in discussions and in developing strategies and management plans.
Designing Foodprint to support collaboration
We designed Foodprint to support two different health contexts: irritable bowel syndrome (IBS) and healthy eating. In both contexts, people use food diaries and work with health experts to support their goals. People with IBS track their food and symptoms to figure out their symptom triggers. They often work with clinicians, nurses, or dietitians to develop individualized diagnosis and treatment plans. People with healthy eating goals often track a variety of information, such as food, exercise, or sleep to find opportunities to improve their health and wellness. Many people also share the data with nutritionists or health coaches to monitor the progress.
In earlier research, we found that people are often trying to use personal tracking data as Boundary Negotiating Artifacts in collaboration with and among health experts.
Boundary negotiating artifacts are artifacts used in the coordinating and negotiating process across parties in collaborations. There are five types of boundary negotiating artifacts:
- Self-explanation artifacts: These artifacts are created for personal use. We designed Foodprint mobile app that people use to keep track of their diet as a self-explanation artifact.
- Inclusion artifacts: These artifacts are created to present new concepts to support collaboration. For example, we designed the photo-based visualizations in the Footprint web app to show relationships between food and individual health goals.
- Compilation artifacts: These artifacts help develop shared understandings across multiple groups. One example would be the notes in the electronic medical record (EMR), which can include patient medical history, test results, and recommendations.
- Structuring artifacts: These artifacts help direct and coordinate activities. Notes or summaries printed from EMR are examples of structuring artifacts.
- Borrowed artifacts: These artifacts are usually augmented from other types of artifacts and adopted in unanticipated ways.
While designing Foodprint, we drew on these previous results and the different categories of boundary negotiating artifacts to plan its capabilities and designing its interface.
The resulting system includes three components: (1) a mobile app that allow people to record what they eat using photos, (2) a web app that presents visual summaries to show relationships between food and individual health goals, and (3) pre-visit notes that prompt people to reflect on their data and record their goals and questions before meeting health experts. In this way, we sought to support self-explanation for individual use, inclusion and compilation for collaboration with health experts, and structuring for planning what steps to take after a clinic visit.
Using the mobile app, people can take a photo whenever they eat. For people with IBS, we added a few questions to help identify ingredients that might be symptom triggers but are difficult to see, such as spice, dairy products added to beverages, or dressings. We also asked them to track their symptoms in a frequency they defined (Figure 1). People with healthy eating goals could focus on one of three common healthy eating goals: balanced diet, monitoring specific ingredients (e.g., sugar), and observing relationships between food and emotions (Figure 2). While people commonly have these goals, current commercial applications rarely support them. Instead, they tend to focus on weight management or calorie goals.
The web app presents the relationship between food and individual health goals. For people with IBS, the web app presents food photos based on symptom severity and time. People can also choose to categorize their foods based on food source and preparation type. These categorizations allow individuals and health experts to explore potential triggers based on individual experience (Figure 3). For people with healthy eating goals, the web app presents food photos based on their chosen health goals (Figure 4).
We designed pre-visit notes to support people to reflect on their data and include their individual goals and experience into the discussion with health experts. We asked people to summarize their findings from the tracked data, their goals for the upcoming visit, and questions they would like to ask health experts in the visit (Figure 5). We then shared the pre-visit notes with the health experts before the visit.
How do people use Foodprint with health experts?
We deployed Foodprint to 33 participants and 16 health experts. 17 participants with healthy eating goals used Foodprint to record their food for around a month (on average 21 days, SD=7.6 days) and then met with a health expert for a dietary consultation. 16 participants with IBS tracked 12 days on average (SD=9.3 days), following current clinical protocols, and then met with an IBS provider for an IBS consultation. During these visits, we observed how Foodprint enable individuals and experts to share knowledge to support data interpretation and incorporate these discussions into actionable plans. We share examples from our study that illustrate the variety of ways people and their health experts used Foodprint in collaboration.
Photo-based visualizations provided context and patterns that support data interpretation
One participant had IBS for over 17 years but still could not pinpoint the triggers for her symptoms. When she reviewed the Foodprint visual summary with the provider (Figure 6), each noticed that the same foods appeared in all symptom columns, which could indicate that these foods are not potential triggers. However, the provider also observed that almost all the photos had a steering wheel in the background, which indicated the participant’s busy work routine. They then had a conversation about how stress and the busy schedule might have exacerbated symptoms. Consequently, rather than focusing on eliminating specific foods, they then focused on strategies that could mitigate stress and eating on the go.
Photo-based visualizations enabled people and health experts to exchange knowledge and experience that help individualized diagnosis
Besides observing trends and patterns, participants and health experts also referred to photos to exchange specific knowledge or strategies that support health goals. For example, one IBS provider noticed that a participant had a very restrictive diet but was able to tolerate nuts (a common IBS trigger food) (Figure 7). The participant explained that she found soaking the nuts helped with digestion. The provider then went on to discuss how soaking helps leach out some of the carbohydrates (FODMAP, Fermentable Oligo-, Di-, Mono-saccharides, And Polyols) that potentially trigger IBS symptoms. The provider also provided other similar examples which the participant could try to incorporate into her diet, such as eating firm tofu rather than silken tofu.
People and health experts used photos to support individualized, actionable plan development
Participants and health experts also used photos as examples of meals that participants could replicate later. One participant found that big meals often upset his digestive systems, and so he and the dietitian talked about adopting smaller, more frequent meals. They discussed whether the participant’s work schedule could accommodate these shorter meals, using the participant’s photos of small yet nutritionally balanced meals as examples to follow.
Design principles from this work
The findings from our research show that personal tracking system designs inspired by boundary negotiating artifacts can help individuals communicate their experiences and elicit knowledge from health experts. Reflecting on the process, here are some thoughts on how designs can support individuals and health experts collaborate further.
- Supporting individual reflection to strengthen collaborative reflection
We designed the Foodprint mobile app as a self-explanation artifact. Our design focuses data collection on what participants should track, based on their individual health goals. Also, although the Foodprint web app was designed as an inclusion artifact to support collaboration, participants had access to photo-based visualizations without health expert involvement. We found these design decisions not only help people learn and reflect on what they eat on their own but also helps them explicitly incorporate these reflections into conversations with health experts. Future systems could further support individual reflection by proposing reflective prompts or questions. These systems can also record individual goals, summaries, and questions from the pre-visit notes to present a history of hypothesis development, verification, and behavior modifications to help individuals reflect on their experiences and choices.
- A shared patient and provider view supported co-interaction and co-interpretation
Designing the photo-based summaries and pre-visit notes as inclusion artifacts supported people and health experts in collaborative reviews. In particular, having access to photo-based summaries and being prompted to reflect on their data by the pre-visit notes allow participants in our study took an active role in interpretation and collaboration. However, technologies that can support co-interaction opportunities with the visualizations can further support patient-involvement in the collaboration. For example, there are more and more remote dietary consultation services that intend to improve access to expert resource today. However, using current screen-sharing technologies, people would not be able to interact with the data when health experts share their screens. Allowing both parties in remote consultations to simultaneously interact with a shared view could help participants and health experts be even more engaged in the co-interpretation process.
- Integrating tools to support the implementation of post-visit plans
Although we did not specifically design for the structuring artifacts, we found people and health experts use photos and the visualizations to structure potential plans after visits. While using the photos as examples was a useful strategy for our participants to replicate their meals, people and health experts can use more help to document these actionable plans and identify the right tools to support these plans. Incorporating post-visit notes into the system and use it to configure other tools that further support hypothesis formation and testing, such as TummyTrials, can help people focus on more directed examination.
Considering a more automated future
When we discuss this project with others, many suggest incorporating automated filtering or analysis features to support data interpretation. While these features might save time, we caution designers against focusing so much on quantification that they overlook the value of contextual, qualitative information. For example, the contextual information shown in the above examples, revealing that patients eat in the car every day, allows health experts to better understand the patients, identify individualized strategies, and provide affectional support. Also, showing and reviewing food as meals, instead of caloric or nutrient information, encourages people and health experts to have a conversation about people’s everyday life and their individual goals, priorities, values, and constraints. This in turns helps health experts develop understandings and empathy as well as focus on specific action plans that are consistent with individual goals, preferences, and constraints.
Deciding what analysis should be automated, as well as who should engage in different integration and reflection activities, when data from personal informatics systems are collaboratively reviewed is an ongoing research and design challenge. As a community, we still have a lot to learn about the tradeoffs of design decisions between increasing efficiency and encouraging individual and collaborative reflection.
To read more, please see our paper from the Proceedings of the ACM on Interactive, Mobile, Wearable and Ubiquitous Technologies (IMWUT). If you want to know more about the research or share your thoughts, we would love to hear from you! Please contact Christina Chung at firstname.lastname@example.org.
Acknowledgment of funding
This research was funded in part by the Intel Science and Technology Center for Pervasive Computing, the Agency for Healthcare Research and Quality (project #1R21HS023654), the National Science Foundation (project #s OAI- 1028195, IIS-1344613, IIS-1553167, and IIS-1813675), and a UW Innovation Research Award. This research was supported by the National Center For Advancing Translational Sciences of the National Institutes of Health under Award Number UL1 TR002319. The content is solely the responsibility of the authors and does not necessarily represent the official views of funding agencies.