The countable symptom beats the subtype label for rosacea tracking.

The label froze in 2019. The rosacea didn't.

A patient walks into a dermatology appointment in 2026 still carrying a label from 2019: Subtype 2.

The chart says papulopustular, the bumps-and-pustules variant. But the papules (inflammatory bumps) mostly cleared two years ago. What flares now is a hot, stinging face after coffee and a left eye that turns gritty by mid-afternoon. On paper, nothing moved. Underneath the paper, almost everything did.

And every visit still opens from the frozen category instead of from what actually changed this month. That's the gap we want to close, and it has a simple fix that has nothing to do with picking a better label.

Track events, not categories

The fix is mechanical. A subtype is a category you were sorted into once, for the purpose of diagnosis and clinician-to-clinician shorthand. It's built to hold still. The thing you actually experience does not hold still.

So the unit you track has to be something that can be different by next Thursday. A flushing episode (a transient bout of warmth and redness) has a start time and an end. A papule can be counted. An eye-irritation day either happened or it didn't. Warmth after a trigger lasts a measurable number of minutes.

We call these countable units: symptom-specific, time-stamped, and small enough to log honestly on a bad day. Each one moves on its own timeline. That independence is the whole point, and it's exactly what a label or a single score erases.

The subtype is a snapshot for the chart. The unit you track has to be something that can change by Thursday.

Why the subtype and the score both fail as tracking units

Subtypes were never designed for longitudinal self-tracking. The original 2002 system (Wilkin et al., JAAD) sorted rosacea into four subtypes to standardize diagnosis and research populations. Useful for that. Useless for capturing that your redness is quieter this month but your eyes are worse.

Then dermatology itself moved on. In 2017 the field shifted from the subtype model to a phenotype approach, describing rosacea by its individual features instead of a single bucket. The chart label many patients still carry predates that shift by years.

Composite severity scores have the opposite problem. They sum graded features into one number, and a single number can sit flat while your week swings hard: papules down, flushing up, the total unchanged. You feel the change. The score doesn't show it.

What the RASI validity literature actually shows

The case for splitting the score apart, rather than trusting the total, starts with the structure of the score itself. The Rosacea Area and Severity Index (RASI) is a composite that grades several features and sums them into a single number. Those features do not all move the same way. Some track with how a patient actually feels. Others move independently of patient experience. A single summed number hides that split by design. If part of the score tracks your quality of life and part of it doesn't, then the part that does is the part worth logging on its own, as its own countable unit, on its own line.

A flare week, converted

Take the Subtype 2 patient and drop the label. Log the week as units instead.

Monday: two flushing episodes, the longer one about forty minutes after a hot shower. Papule count: four, from the morning photo. Eye-irritation day: yes. Wednesday: no flushing, papules down to two, eyes fine. Saturday after a glass of red wine: one long flushing episode, warmth lingering close to an hour, eyes gritty again by evening.

None of that fits inside 'Subtype 2.' All of it fits inside four countable lines. And now a pattern is legible that the label could never hold: the eyes and the flushing track the triggers together, while the papules drift on a slower clock of their own.

What changes when the unit is right

The dermatology visit stops opening from a category and starts opening from a trend. You can see whether a treatment moved your eye-irritation days even in a month when the redness barely budged. You can tell a genuine flare from a bad photo day.

It matters most for the people the visible-redness framing serves worst. On deeper skin tones, erythema is harder to see and easier for a clinician (or an algorithm) to miss. Adamson & Smith 2018 documented that deployed dermatology tools fail harder on dark skin, and Daneshjou 2022 found most published derm image datasets barely represent Fitzpatrick V and VI at all. Sensory units (burning, stinging, warmth duration, gritty eyes) don't depend on how legible your redness is to a camera. They're countable for everyone.

Building the log around the unit

We built Skinframe by starting from that literature and working backward to what belongs in a log. Countable, time-stamped symptoms, each on its own line. Photos as dated evidence, kept on your device rather than shipped to a scoring server. No subtype gate on the way in, and no single index blurring the four things you came to track.

What we deliberately left out is as telling as what we kept. No skin-scanning classifier trying to guess your severity from a selfie, for exactly the reasons Adamson & Smith and Daneshjou describe. The job is to make the trend real, not to hand you a verdict. It's a tracking tool, not a diagnosis.

What we're watching next

Two things. Whether the RASI component-versus-quality-of-life split holds up in follow-up validation, because that's the strongest published case for tracking features separately rather than as a sum. And whether the 2017 phenotype standard finally reaches the intake forms, so a newly diagnosed patient in 2027 is asked what changed this month instead of which bucket they belong in.