Glass Reverie

Good data that nobody had time to use

Consumer health devices can capture a remarkable amount of physiological data. The gap was never collection — it was translation. Clinicians had minutes, not hours, and a stream of raw patient data created more work rather than more clarity.

Meanwhile patients and caregivers were asked to track diligently without seeing how their effort connected to their care. Both sides of the loop were doing work that felt disconnected from outcomes.

Design question

How do we design a health experience that fits inside a real clinical workflow — where the clinician's attention is the scarcest resource in the system?

How I approached it

Four user groups, one workflow

I ran end-to-end research with clinicians, operations staff, patients, and caregivers. Shadowing sessions mattered most — watching a clinician move through a patient visit revealed constraints that no interview surfaced, particularly how little uninterrupted screen time actually exists.

What I learned

• Clinician attention is the bottleneck. Anything requiring more than a glance would not survive a real visit.
• Trends beat readings. A single data point is noise; direction of change is what informs a clinical conversation.
• Caregivers were the invisible users. They often did the actual tracking but had no designed role in the system.
• Operations staff shaped adoption. If enrollment and setup were burdensome, the concept never reached a patient.

Design for the glance, not the deep dive

The team's initial concepts assumed clinicians would explore data. Research showed the opposite: the realistic interaction is a few seconds during a visit. Designing for exploration was designing for a moment that does not exist.

That shifted the work fundamentally — the product's job became interpretation, not visualization. Surface what changed and why it matters, with detail available on demand rather than presented by default.

Concepts tested with clinical partners

Working with clinical collaborators, I explored several models for how patient-generated data could enter a visit: a full dashboard, a pre-visit summary, and an exception-based alert model that surfaced only meaningful deviations.

Testing what we chose

The pre-visit summary won because it respected existing workflow rather than interrupting it — clinicians already prepare briefly before a visit, so the summary occupied a moment that already existed instead of asking for a new one.

Closing the loop for both sides

Pre-visit summary for clinicians

A single view answering three questions: what changed, is it clinically meaningful, and what should we discuss. Full data remains one tap away but is never the default.

Patient tracking with visible purpose

Patients and caregivers see how their tracking connects to their care plan, which reframed logging from an obligation into participation.

Caregiver as a designed role

Rather than treating caregivers as an edge case, the flow gave them an explicit role with appropriate permissions — matching how care actually happens at home.

What changed

• Two concepts advanced to clinical pilot with institutional partners.
• Reframing from visualization to interpretation became the team's shared design principle for later projects.
• Caregiver needs were formally added to the product's user model.
• Research artifacts were reused across multiple projects in the digital health portfolio.

Note: specifics are generalized; clinical and product details are confidential.

What I'd do differently

I spent early effort designing rich data views before I understood the clinical workflow well enough. Shadowing sessions should have come before concepting, not alongside it — they invalidated assumptions I had already built on.

Working in a clinical context taught me to design for the constraint rather than the ideal. The best solution here was the one that asked the least of the person with the least time.