Results

The situation

The dashboard already existed. Remote monitoring did not.

Heal sent doctors to patients' homes for same day care. The next step was turning those episodic visits into continuous primary care.

For patients over 65 with hypertension or diabetes, the moments that matter happen between visits. A reading that drifts. A medication that stops working.

Remote monitoring did not exist. Heal decided to prescribe Bluetooth devices, connected through Wellbe, to patients with chronic conditions, and there was no product for what a doctor would do once that data started arriving.

So my job was to define and build that product. Not a screen, the whole path from a device reading to a clinical decision, from zero.

The 0 to 1 problem

Build a new care model.

Move Heal from episodic home visits to continuous primary care. Success meant repeat patient rate at 39% to 45%, Medicare approval, and clinicians able to act on monitoring without a second workflow.

Constraints

• Clinical safety. A missed threshold is a patient outcome, not a UX bug.
• One month. Validated with practicing physicians inside a single month.
• Reimbursement. The workflow had to satisfy Medicare's chronic care rules by default.

Discovery

One month meant building the research cadence into the sprint, not running discovery before it.

I started with an interview with Heal's Head of Medicine: how and when would a doctor recommend monitoring, and which conditions should we target first.

With engineering I mapped the device ecosystem from zero: which Bluetooth devices to support, how a reading gets from a patient's home through Wellbe to the patient’s phone to Apple Health to the Heal cloud.

Which measures, and why

We selected against HEDIS, the measure set more than 90 percent of US health plans use to score care quality.

From data stream to decision system

My focus was to turn a continuous stream of data into an ordered set of decisions a doctor could finish between house calls.

Explorations

I started with monitoring at the patient level. What I needed was a triage flow.

My first version put charts on the patient screen, based on what we heard in discovery. The panel liked the visualization, but quickly showed us that this wasn’t how they needed to work through a patient.

A doctor between house calls does not open a chart to explore. They need to know who needs them today, then finish a decision and close the loop.

Exploration · Patient-level monitoring

What I built

What mattered most was the daily overview: “What do I need to tackle today?”

1. Daily Overview. The screen doctors already open first. Monitoring alerts sit inline beside today's visits and unsigned charts, ordered by what needs attention rather than by when it arrived.
2. Workflow to Review. From an alert, a doctor reviews the signal, chooses an outcome, and closes the loop without leaving the day’s work.
3. The alert logic. Not every abnormal reading is an alert. Not every alert belongs with a doctor. Not every missing reading means danger. Defining that was the actual product problem, and the artifact that let this scale without me in the room.

Closing the loop

Nothing gets left unresolved. Every item has a clear outcome.

In Review

Take action, or close the alert. Either way, it leaves a record.

• Change a prescription
• Book an appointment
• Contact the patient today
• Acknowledge it and take no action
• Request device support

What shipped

Daily Overview, then a Review workflow that turns a signal into an outcome.

Open the after prototype →

The results

We did not just redesign a view. We built the workflow Heal needed to make monitoring part of primary care.

39% fewer unnecessary prescriptions for monitored patients. The clearest evidence the workflow changed clinical reasoning and not just the screen.

39% to 47% repeat patient rate, above the 45 percent goal, while acute episodes fell.

Medicare approved. Monitoring became free to patients and made home primary care economically viable.

Over $4M a year in recurring net margin.