Designing tools and practices for patients and clinicians to make shared use of patient-generated health data.
People generate health data between clinical visits: food diaries, symptom logs, blood pressure readings, activity records, and medication logs. Most consumer health apps, however, are designed for personal use, not for sharing with a care team. Patients who want to bring their data into clinical conversations often find that neither their tools nor clinical workflows support it, and clinicians who receive patient-generated health data (PGHD) often lack the time, context, or interfaces to use it well.
In projects funded by the Agency for Healthcare Research and Quality (1R21HS023654), the Patient-Centered Outcomes Research Institute (PCORI), and the National Library of Medicine (1R01LM012810), my collaborators and I study how patients and clinicians collaborate around personal data, co-design tools to support that collaboration, and identify the conditions under which PGHD improves care. This work spans multiple conditions, populations, and clinical settings, with a focus on chronic conditions where ongoing self-monitoring is clinically important and burdensome for patients.
Before designing tools to improve data sharing, we needed to understand current practices and why they so often break down. Working with patients managing irritable bowel syndrome (IBS) and overweight or obesity, two conditions where clinicians often recommend self-monitoring, we found that even when patients brought detailed records to clinical visits, providers often did not use them in the ways patients hoped or expected. Providers faced time constraints, lacked visualizations suited to clinical interpretation, and were uncertain how to act on data collected with consumer tools. The barrier was a mismatch between what tracking tools produced and what clinical encounters could support.
In a systematic review, we synthesized evidence on how PGHD and patient-reported outcomes affect the patient-clinician relationship across healthcare settings, and identified consistent changes to communication, trust, and clinical authority that have implications for tool design.
In a PCORI-funded project, we examined opportunities for PGHD to transform healthcare more broadly, through interviews and surveys across a range of clinical settings and patient populations. This work documented the diversity of data types, clinical use cases, and implementation barriers, and produced a framework for characterizing the stakeholder perspectives that shape how health systems adopt PGHD.
Christina Chung led our study of people managing IBS or weight who tracked food and symptoms with consumer apps and other tools. Drawing on Charlotte Lee's concept of boundary negotiating artifacts, objects people use to negotiate the boundaries of their expertise, autonomy, and responsibility, we found that tracking data prompted patients and providers to renegotiate who controls what, whose knowledge counts, and how they make clinical decisions. This framing treats data sharing as a social negotiation rather than an information transfer, and it suggests how tools should present data at the clinical boundary.
These findings pointed to a design opportunity: instead of asking patients to bring raw tracking data to visits and hoping providers could interpret it, tools could organize data around the questions clinical conversations address. Two lines of design work built on this insight.
FoodPrint. Studies of IBS food and symptom journals showed that patients and providers struggled to find meaningful patterns together in detailed records. FoodPrint offers lightweight food diaries organized around the questions patients and providers need to answer, helping them identify patterns, set individualized goals, and plan concrete changes together.
TummyTrials. Once a patient and provider suspect a trigger, how can they test it? Instead of sifting through accumulated data and hoping patterns emerge, often confounded by other factors, TummyTrials helps patients form specific hypotheses about food triggers and run brief self-experiments whose results both patient and provider can interpret.
FoodPrint and TummyTrials each addressed a specific clinical context. Susanne Kirchner led our work on what clinical use requires of personal health informatics tools more generally. We identified translational needs: design requirements that arise at the boundary between personal tracking and clinical use, and that consumer tools often fail to meet. The concept gives researchers and designers a vocabulary for building tools that bridge personal and clinical data practices.
Most fitness tracking happens outside the clinical loop: data that could inform care decisions sits in consumer apps that providers never see. The ExerciseRx program, developed with The Sports Institute at UW Medicine and the UW Ubicomp Lab, aims to close that gap. The platform integrates patient-facing mobile tracking with clinical dashboards in the EHR and provider-patient messaging, so that exercise data collected at home becomes part of the clinical conversation. Current deployments span oncology prehabilitation and rehabilitation, multiple sclerosis, and cerebral palsy, contexts where home-based exercise is clinically important but has been hard to monitor and support between visits.
Guidelines recommend confirming elevated clinic blood pressure with measurements taken outside the clinic, but in practice clinicians rarely order ambulatory monitoring and often rely on follow-up clinic visits. In BP-CHECK, a PCORI-funded randomized diagnostic study of 510 primary care patients at Kaiser Permanente Washington (formerly Group Health), we compared clinic, home, and kiosk blood pressure measurements against 24-hour ambulatory monitoring, the reference standard. Home monitoring performed best: its average readings nearly matched ambulatory monitoring, and patients preferred it. Follow-up clinic measurements, as taken in routine care, read lower than ambulatory monitoring and missed about two-thirds of people with hypertension. Kiosk measurements read higher and produced more false positives. Even after ambulatory monitoring confirmed hypertension and the results went to their primary care providers, only 40% of these patients had a diagnosis recorded within six months.
Surveys and interviews with physicians, nurses, and medical assistants helped explain the gap: many trusted manual clinic measurements most, few knew the thresholds for diagnosing hypertension from home or ambulatory readings, and they lacked ways to bring home readings into the EHR. This work built on our earlier study of a blood pressure kiosk in a family medicine clinic and has since extended to email outreach that helps more patients measure at home. The full PCORI report is freely available.
Sharing personal data can reinforce or challenge power asymmetries beyond the patient-clinician relationship. In college athletics, for example, institutions deploy wearables on athletes, raising questions about consent, data ownership, and the difference between data collected for an athlete and data collected about one. I discuss this work alongside related projects on power, equity, and agency in technology design.