What did the numbers show?

The AI discharged a substantial share of patients entirely on its own. Autonomous discharge rates were 37% at site 1 and 32% at site 2. After exclusions, actual discharge rates were 31% and 25%, respectively. Teledermatologists then discharged an additional 24% and 25% of patients at the two sites.

The autonomous pathway also reduced two forms of medical intervention. Routine follow-up dropped from 27% under standard teledermatology to 12% with the autonomous AI. Biopsy rates fell from 43% under conventional face-to-face care to between 26% and 29%.

False negatives — cases the AI wrongly labeled benign — were rare, and within the available follow-up period the researchers identified no adverse outcomes among patients discharged autonomously.

What are the limitations?

The findings come from a conference presentation rather than a peer-reviewed publication, so they have not yet undergone formal external scrutiny. The follow-up window, while showing no harm, is described only as "available follow-up," and longer-term confirmation of safety would strengthen the case for autonomous discharge.

The study also covered two hospital sites within a single health system. Whether similar consent rates, discharge rates, and safety results would hold across other clinics, countries, and patient populations remains an open question. An 86% consent rate is high, but it also means roughly one in seven patients declined autonomous decision-making — a reminder that patient acceptance is itself a constraint on deployment.

Why capacity matters more than the technology

For Thomas, the central point is not the algorithm but what it frees specialists to do. "We believe the greatest value of autonomous AI lies not in the technology itself, but in the specialist capacity it unlocks," Thomas said in a statement.

That framing matters for dermatology services under pressure. Urgent skin cancer pathways typically funnel every referral — the vast majority of which turn out to be benign — through scarce specialist time. A system that can safely discharge roughly a quarter to a third of patients without a clinician touching the case shifts that specialist time toward the patients who need it most.

The 62% capacity gain is an estimate, and it applies to the specific pathways studied. Still, at a time when many health systems face dermatologist shortages and growing referral volumes, the direction of the result is notable: carefully validated, risk-stratified AI did not replace clinicians but redistributed their workload.

What happens next?

The study adds to a growing body of evidence on AI triage in skin cancer detection, though autonomous discharge — sending patients home with no human review at all — remains the most ambitious and contentious step in that field. Longer follow-up, independent replication at additional sites, and peer-reviewed publication will determine whether the Vienna results hold up.

For now, the two-hospital experience suggests that, with high patient consent and careful safeguards, autonomous AI can take a meaningful slice of routine diagnostic work off dermatologists' plates — 2,851 hours' worth, by the researchers' estimate — while flagging the high-risk cases that genuinely need a specialist's eyes.