Dermatologists want a one-page summary. They don't want your dashboard URL.

The 12-minute visit economy

A dermatology follow-up visit in US clinical practice runs roughly 12 to 15 minutes. The dermatologist has read your chart in the 30 seconds before walking in. The first three minutes are spent on listening to what changed, the next five on the focused exam, the next three on plan and prescription, and the closing minute on logistics. There is no slack in that schedule for a 90-photo album.

What the dermatologist wants out of your tracking is whatever moves the diagnostic and treatment-planning needle inside that window. The format that does this, per the teledermatology and patient-reported-outcome literature, is a single-page or 2-page printable summary the patient brings in. The dermatologist scans it in the first three minutes, asks two or three follow-ups, and uses it to anchor the plan.

What the dermatologist does not want is a portal integration that requires logging into a separate system, a dashboard URL on the patient's phone the clinician has to read sideways while the patient holds it, or a wall of daily photos with no annotation. The literature is explicit about this; the consumer-app category has been slow to read it.

What the published preferences identify as high-value

The dermatology PROM (patient-reported outcome measures) literature has been working out what makes a patient-tracking handoff clinically useful for at least a decade. The high-value signals consistently identified in the published evidence on dermatology-app handoffs (see Naik et al. JAAD 2022 on PROMs as a vital sign; the systematic review and meta-analysis on atopic dermatitis apps, Therapeutic Advances in Dermatology 2024; the broader teledermatology review literature):

- Frequency and duration of flares over the visit window. A count of flare days, average duration, longest flare. A number, not a sentence. - Identifiable triggers with simple co-occurrence counts. "Out of 12 flares this period, 8 followed sun exposure within 24 hours, 6 followed alcohol within 12 hours." Co-occurrence with explicit framing that it is not causation; the dermatologist does the causal inference, the tracker provides the counts. - Severity trajectory. A per-feature trajectory, not a single global score (see our companion piece on severity scoring). The dermatologist wants to see whether the redness is moving, whether the papules are moving, whether the ocular component is moving, each on its own line. - Two photos. One representative "good day" and one representative "bad day" from the visit window, with timestamps. The literature is explicit that the dermatologist does not want a daily-photo dump; they want two photos that frame the range. The good day plus the bad day is what the chart actually needs. - Treatment adherence summary if the patient is on a prescription. Days the patient applied, days they missed, why if known. The published evidence is clear that misunderstood adherence is one of the most common reasons a treatment looks like it failed when it was not fully tried. - Quality of life trend, quantified. The RosaQoL or DLQI or similar instrument's score across visits, ideally as a single number with the previous-visit comparison shown next to it. The Naik et al. JAAD 2022 piece is the canonical citation for PROMs as the dermatology vital sign.

None of those signals require a model or an inference layer. They require the patient to record the data and the tracker to summarize it cleanly. The complexity is in the design of the summary, not in any computational machinery behind it.

What the literature flags as noise

The same body of evidence identifies what dermatologists treat as noise and what actively erodes trust:

- Daily photo dumps with no annotation. The dermatologist cannot interpret 60 timestamped selfies in 30 seconds. The information density is wrong for the visit. - Trigger logs without severity context. Knowing the patient ate tomatoes on Tuesday does not help the dermatologist if the tracker has not also recorded whether the rosacea was active on Wednesday. The trigger and the response need to be in the same record. - Subjective free-text journals without structure. The dermatologist will not read three paragraphs of narrative under time pressure. A structured field with a number is read; a paragraph is skipped. - Pseudo-diagnostic scores from consumer apps. When the tracker outputs "your skin score is 7.2 out of 10" and the patient asks the dermatologist what to make of it, the dermatologist is being asked to validate or override a number whose computation they cannot inspect. The most common response is a quiet decision to treat the tracker output as decorative. Repeated, this erodes the dermatologist's trust in patient-tracking generally, which is worse than no tracker at all. - Portal integrations the dermatologist's EHR cannot accept. The teledermatology review literature is explicit that most dermatologists are not on the same EHR as their patients' trackers. Pushing FHIR-formatted data into an integration the practice cannot read is no better than pushing nothing.

The pattern across the noise list is the same: the noise items either ask the dermatologist to do work the visit does not allow time for, or ask them to validate a layer of inference they did not produce.

Why PDF, not portal

The teledermatology and consumer-tracking literature converges on a perhaps-surprising shape for the handoff: a structured printable PDF or printed summary, brought to the visit by the patient, scanned during the 30-second chart review and the first three minutes of the visit (see the teledermatology review at PMC 2023, and the PROM application surveys).

The argument for PDF over portal integration is practical. Most dermatology practices use one of a handful of EHRs (Epic, NextGen, Modernizing Medicine, athenahealth in dermatology). None of them has a universal patient-tracking integration. A FHIR push from a consumer app to the dermatologist's EHR is technically possible but operationally rare. A FHIR push that requires the dermatologist to look at a second screen during the visit is operationally rejected. The PDF, in contrast, the patient hands the dermatologist on paper or on the phone, the dermatologist looks at it, the visit proceeds.

The practical hierarchy the literature supports is therefore:

- PDF first. The simplest format that crosses every clinic. Should be one or at most two pages. - CSV second, for the patient who wants the data themselves. Always free, never paywalled; the patient owns their tracking data. - FHIR third, for the practices that can ingest it. Not the default; offered to the small share of clinics that have asked for it.

Most consumer trackers ship the dashboard-URL or in-app share as the primary handoff. The literature predicts that this is the format the dermatologist will work with the least often. Skinframe ships the PDF as the headline handoff, the CSV as the available alternative, and treats FHIR as a future surface for the practices that ask.

How a one-page handoff is shaped

There is no formal industry standard for the rosacea-tracking handoff PDF. The standard is implicit in the high-value-signal list above. A handoff that hits that list on a single page is the right shape.

The page Skinframe ships is structured as follows:

- Top band, three numbers. Flare-day count for the visit window. Average per-feature severity. Quality of life score (RosaQoL or a translated equivalent), with the prior-visit comparison if available. Three numbers, set in large type, dermatologist-readable in two seconds. - Middle band, trigger co-occurrence. Top three triggers by co-occurrence with flare onset, with the explicit "co-occurrence, not causation" footnote. The dermatologist does the causal inference; the tracker provides the counts. Each trigger row carries an evidence-tier label (well-established / patient-reported / individual-discovery) so the dermatologist knows whether they are looking at Tier A literature triggers or Tier C user-defined ones. - Photo band, two photos. One representative good-day photo, one representative bad-day photo, both timestamped. White-balance-locked, no flash, fixed face geometry; the methodology footer notes the capture protocol so a clinician knows the lighting is comparable between the two shots. - Methodology footer. The phenotype framework Skinframe is built on (ROSCO 2017, NRS 2017), the per-feature severity descriptors (PSA-style), and a note that the tracker is not a medical device, does not diagnose, and complements rather than replaces clinical assessment.

This fits on US Letter. It scans in under 30 seconds. It carries the information the literature identifies as high-value, in the format the literature identifies as high-uptake. There is no portal integration, no model output, no "diagnostic confidence" number. The patient owns the file; the dermatologist scans it; the visit proceeds.

The honest evidence gap

We should be straight about one thing. There is no published rosacea-specific randomized controlled trial showing that patient-tracking-app use improves dermatology visit outcomes. To the best of our reading, no such study has been done. Most consumer-tracker apps that imply otherwise are drawing on adjacent dermatology conditions, especially atopic dermatitis.

In atopic dermatitis, the evidence is real. A 2024 systematic review and meta-analysis of mobile self-management interventions across 1,038 patients showed a mean POEM improvement of negative 1.57, a clinically meaningful effect, with DLQI improvement alongside (Therapeutic Advances in Dermatology 2024). The Naik et al. JAAD 2022 piece frames patient-reported outcome measures as the dermatology "vital sign" with the supporting argument that PROMs help patients communicate concerns rapidly and systematically, increase efficiency, improve the clinician-patient relationship, and increase patient satisfaction.

It is reasonable to think the same effect applies to rosacea, where the disease structure is similar (chronic, relapsing, trigger-responsive, quality-of-life-affecting). The adjacent evidence is the honest framing we use: "evidence in adjacent dermatology conditions such as atopic dermatitis shows that patient-reported tracking improves visit outcomes; Skinframe applies the same approach to rosacea." That is what is true.

What we do not say is that rosacea-tracking has rosacea-specific trial evidence behind it, because the rosacea-specific trial has not been run. The literature on adjacent conditions is the bridge we use; the bridge is honest as long as we name it as a bridge.