"Abandonment" was the framing the field used to describe people stopping their use of tracking tools — a framing that treats non-use as failure. Our research challenged this: through interviews with people who had stopped using fitness trackers and other apps, we found that lapsing is often intentional, often temporary, and often accompanied by unmet needs — for tools that help people return gracefully, maintain insight across gaps, and transition between different modes of engagement.

Collecting data is not the same as understanding it. Early work on lifelogs examined how visualizations of personal data can be designed to support sensemaking — developing the concept of cuts (subsets of data with a shared feature) as a way to surface actionable patterns rather than overwhelming users with raw quantity.

The shift from population-level evidence to individualized care is one of the central challenges of modern medicine — and personal tracking data is one of the most promising resources for making that shift. But making good use of tracking data in the service of personal health goals requires a different study design than standard clinical trials. This paper argues for the importance of starting with goals — the patient's goals, not the researcher's or the clinician's — in designing and interpreting N-of-1 (single-subject) studies that use tracking data.

For people managing chronic health conditions — where tracking is often recommended but rarely sustained — the gap between what tracking tools offer and what patients need is especially consequential. A multi-method study with people managing migraine and other chronic conditions examined how people's goals for tracking evolve, what kinds of data they find meaningful, and where current tools fall short. This work produced design recommendations grounded in the realities of long-term tracking under conditions of uncertainty and symptom variability.

Goal-directed tracking can support two distinct kinds of investigative work, which require different tool designs and different kinds of effort from trackers.

Foodprint structured lightweight food diary data to help people and their care teams identify candidate patterns — what to examine more closely, what might be worth testing. The tracker's job is to log carefully and engage with what the structure surfaces; the tool's job is to make that collaborative sensemaking tractable rather than overwhelming.

TummyTrials picked up where pattern discovery leaves off: once a person suspects a trigger, how do you actually test it? The system supported people with IBS in formulating specific hypotheses about food triggers and designing brief personal experiments — a different and more demanding kind of tracking work, but one that can produce individualized evidence neither population studies nor open-ended logging can provide.

Foodprint and TummyTrials also emphasize the importance of collaboration with a care team. See more about my research on integrating Patient-Generated Health Data into Clinical Care →

A study of menstrual tracking apps — among the most widely used health tracking tools globally — found that available apps were poorly aligned with the diverse goals and needs of people who use them. This work documented how apps reflect oversimplified models of the menstrual cycle and the people who track it, surfacing design implications that have influenced both subsequent research and public conversation about reproductive health technology.

Workplace wellness programs represent a distinctive context for self-tracking — one in which tracking is often incentivized, socially visible, and structured around employer rather than employee goals. A study of employees in workplace wellness programs found that people adopt trackers for the incentives but stay (or leave) based on whether the tool fits their existing health practices and values — raising questions about whose goals wellness programs actually serve.

Most personal informatics research assumes the tracker and the primary beneficiary of the data are the same person. Collegiate sports teams complicate this: athletes wear sensors and generate detailed performance and health data, but the goals of the individual student-athlete and the goals of their coaches and institution are not always aligned — and it is rarely the athlete who controls how the data is used. A study of tracking practices across U.S. college sports teams examined how this tension plays out in practice, finding that athletes often experienced tracking as something done to them rather than for them, and that the design of tracking systems reflected and reinforced existing power relationships within teams.

Many tracking tools present data through detailed dashboards that require deliberate attention. This work examined an alternative: glanceable feedback — brief, ambient representations of progress that can be absorbed in seconds. Through a design exploration with physical activity trackers, we characterized the design space of glanceable feedback and the conditions under which brief representations can be informative, motivating, or unhelpfully oversimplified.

As large language models become widely available, people are increasingly turning to AI systems to make sense of their health data — asking questions about their tracking history, planning behavior changes, and reflecting on patterns in ways that earlier interfaces did not support. A recent study examined how people engage with generative AI as a resource for personal health informatics, identifying four distinct engagement modes (planning, tracking, reflecting, and acting) and surfacing design opportunities and tensions specific to health contexts, including concerns about accuracy, privacy, and the appropriate role of AI in consequential health decisions.