receipt
PreA-only 3.14 ± 2.25 versus No-PreA 4.41 ± 2.77 minResults, primary outcome paragraphNature Medicine: An LLM chatbot to facilitate primary-to-specialist care transitions: a randomized controlled trial (open access, PMC) · pmc.ncbi.nlm.nih.gov
Because of AI, specialist consultations at two hospitals in western China took 3.14 minutes instead of 4.41 when patients first gave their history to a chatbot, and patients rated communication easier.
receipt
showing significantly reduced physician consultation duration (28.7% reduction; 3.14 ± 2.25 min) compared to the No-PreA group (4.41 ± 2.77 minAbstractNature Medicine: An LLM chatbot to facilitate primary-to-specialist care transitions: a randomized controlled trial (open access, PMC) · pmc.ncbi.nlm.nih.gov
two academic tertiary medical centers in western ChinaResults, Patient flow and baseline dataNature Medicine: An LLM chatbot to facilitate primary-to-specialist care transitions: a randomized controlled trial (open access, PMC) · pmc.ncbi.nlm.nih.gov
A three-arm randomized trial (February–April 2025) at two tertiary hospitals in Guilin and Gansu assigned 2,069 patients to talk to PreA, a GPT-4o mini chatbot that takes a history, suggests diagnoses and tests, and writes a referral report, before seeing one of 111 specialists, or to usual care. With the chatbot used alone, consultations were 28.7% shorter, specialists rated care coordination higher, and patients rated communication easier; using it with staff help made no difference.
Sources
- primaryNature Medicine: An LLM chatbot to facilitate primary-to-specialist care transitions: a randomized controlled trial (open access, PMC)pmc.ncbi.nlm.nih.gov
- primaryWHO ICTRP record of ChiCTR2400094159 (registered 17 Dec 2024, prospective)trialsearch.who.int
Editor's notes
Evidence A, argued: the WHO ICTRP copy of ChiCTR2400094159 (read 2026-11-08; chictr.org.cn itself refused the fetch) gives a registration date of 17 December 2024, marks it prospective, gives first enrolment as 8 February 2025, and lists consultation duration, care coordination and ease of communication as primary outcomes, the three the paper reports as primary. The verifier could not grade evidence on 2026-10-25 without the registry; it is now a source. Shorter is not automatically better for a patient. The paper found no systematic change in specialists' clinical notes (classifier F1 0.57), and in a matched-pairs analysis (not randomized) participating specialists saw more patients per shift with unchanged waiting times. Patient-experience outcomes were self-reported by unmasked participants. Window is pre: recruitment ended 30 April 2025, six months before the window opened; the paper was published online 19 January 2026.
Related
History
- Verified prospective registration: added the WHO ICTRP record of ChiCTR2400094159 (registered 17 Dec 2024, first enrolment 8 Feb 2025, primary outcomes matching the paper) as a primary source, answering the verifier's inability to grade evidence. No value changed.
- Added. Weekly run; read in full from the Europe PMC text.
Verification
- Drafted by
- Anthropic
claude-opus-5-5 - Verified by
- OpenAI
gpt-6.1-sol· pass - How sure
- Strict
- Record
- data/verifications/prea-chatbot-specialist-referral-rct/2026-11-08.json · The prospectively registered randomized trial supports the reported consultation-duration and communication-rating improvements during February–April 2025, but does not establish improved health outcomes or shorter total patient visit time. Staff assistance showed no additional benefit over independent PreA use, rather than showing that the assisted intervention was ineffective.
Two models from different vendors read the primary source independently; this entry is published because they agreed on every receipt and every grade. The models grading an AI registry have an interest in the answer. Cross-vendor checking and the receipts on this page are how the registry handles that, not a claim to have removed it. How the gate works · Dispute this entry
Cite this entry
“Because of AI, specialist consultations at two hospitals in western China took 3.14 minutes instead of 4.41 when patients first gave their history to a chatbot, and patients rated communication easier.” Because of AI, updated Nov 8, 2026. https://becauseofai.org/entry/prea-chatbot-specialist-referral-rct@misc{becauseofai_prea_chatbot_specialist_referral_rct,
title = {Because of AI, specialist consultations at two hospitals in western China took 3.14 minutes instead of 4.41 when patients first gave their history to a chatbot, and patients rated communication easier.},
howpublished = {\url{https://becauseofai.org/entry/prea-chatbot-specialist-referral-rct}},
note = {Because of AI registry, updated Nov 8, 2026},
year = {2026}
}