Patients will accept AI in dermatology — through their own doctor, not instead of one

Design

Cross-sectional random-sample survey with factor analysis and multiple linear regression

Population

10,013 US veterans surveyed after outpatient dermatology care; 1,881 complete responders

Primary outcome

Willingness to use home teledermatology and AI for skin care, scored 1 to 5

Effect

Mean 3.6 overall; PCP guided by computer 3.8, home teledermatology 3.6, computer-guided self-care 3.5

Teledermatology and image-based algorithms are being built on the assumption that patients want them. This survey tested the assumption. Ten thousand and thirteen US veterans who had received outpatient dermatology care were randomly surveyed; 1,881 answered every willingness question on a one-to-five scale.

Willingness was modest rather than enthusiastic: a mean of 3.6 overall. Factor analysis separated three things patients were being asked to accept, and they were not equally acceptable. Care from a primary care physician guided by a computer scored highest at 3.8. Home-based teledermatology scored 3.6. Computer-guided self-care scored lowest at 3.5. Younger age and confidence in completing medical forms — a proxy for health literacy — predicted willingness consistently; living with someone, financial security and education beyond secondary school predicted it inconsistently. The desire to avoid travel, the benefit most often cited when these services are pitched, contributed least of all.

The design lesson is specific and it cuts against how most of these services are built. Patients are most willing when the algorithm sits behind a clinician they already have, and least willing when it is handed to them to use alone. A service built as a patient-facing app is aiming at the least acceptable configuration. In India, where the referral chain to a dermatologist is long and the general practitioner is the first and often only contact, that is the configuration that would have worked anyway — but the health-literacy finding is a warning that the patients with least access are also the least willing, so a technology-first rollout will widen the gap it was meant to close.

• Build image triage to support the referring GP, not as a standalone patient app.
• Do not sell teledermatology on travel saved — patients rank that lowest.
• Plan for an in-person route for older and lower-literacy patients rather than treating them as adoption laggards.
• Ask patients what they are willing to use before building the pathway, not after.
• Keep a named clinician attached to any algorithm-assisted result the patient sees.

Why it matters

It contradicts the assumption that a patient-facing app is the obvious way to deliver skin care at distance.

Don't overread it

US veterans are older, predominantly male and insured differently from most populations; the absolute scores will not transfer.

The statistics, in plain English

These are mean scores on a five-point scale, and the gaps between them — 3.8, 3.6, 3.5 — are small. The direction is consistent across the three factors, but no confidence intervals are given for the differences, so treat the ranking as more reliable than the size of the gap. The larger limitation is who answered: 1,881 complete responses from 10,013 surveyed is under a fifth, and people who complete a survey about technology are likely to be more comfortable with it than those who do not — so the true population willingness is probably lower than 3.6, not higher.

Read the rest in the app

You have read your two free briefings this month. The app carries all 27 specialties, every morning, free — and this finding is waiting in it.

Get Daily Doctor — free

Scan to keep reading on your phone. No account needed to start.