OpenAI's o1 model correctly diagnosed 67% of emergency room patients using electronic records and nurse notes, outperforming triage doctors who achieved 50-55% accuracy, according to a new study.
Researchers found that OpenAI's o1 reasoning model exceeded the diagnostic accuracy of human triage physicians when given access to patient electronic health records and brief descriptions from nursing staff.
The 67% accuracy rate represents a significant gap over the 50-55% baseline for triage doctors—the medical professionals typically responsible for initial patient assessment in emergency departments. Triage doctors make preliminary diagnoses that determine treatment priority and department routing.
The study suggests AI could substantially improve emergency care workflows. Researchers characterized the findings as marking a "profound change in technology that will reshape medicine."
Accuracy in initial diagnosis directly impacts patient outcomes. Misclassification can delay critical treatment or divert resources inefficiently. The study indicates o1's ability to process complex medical data—combining structured records with unstructured clinical notes—may give it advantages in pattern recognition that human physicians struggle to match under time pressure.
The research used real emergency room data, testing o1 against actual diagnostic outcomes. The model's reasoning capabilities appear particularly suited to medicine, where decisions require weighing multiple evidence types and considering rare conditions.
Key limitations remain unstudied. The research does not address whether AI accuracy would translate to real-world deployment, where factors like liability, integration with existing systems, and workflow disruption matter. Triage doctors also handle non-diagnostic responsibilities like patient communication and triaging volume during peaks.
No timeline for clinical implementation was announced. The findings follow increasing AI integration into healthcare, from radiology analysis to drug discovery. Medical regulators will likely scrutinize any tools that directly replace physician judgment.
The results add to evidence that large language models and reasoning-focused AI may find practical value in domains requiring rapid synthesis of complex information. Emergency medicine, with its high-stakes, time-constrained decisions, represents a potential early-adoption area.
Uber's weekly AI agent requests have grown nearly tenfold since February, yet the company has held spending flat since April after exhausting its entire 2026 AI budget in Q1.
A recent paper shows artificial intelligence often diagnoses and treats patients better than human physicians. The findings are prompting difficult conversations within the medical community about the profession's evolving role.
The Relay Q, launching next year, represents the latest push to establish voice as the primary interface for human-computer interaction, challenging the keyboard's decades-long dominance.
An Anthropic researcher demonstrated automated systems that can identify and correct misaligned behaviors without compromising overall performance. The systems improved on all 10 tested benchmarks measuring specific problematic outputs.