aceCan AI assist in reducing diagnostic error? A narrative review
Diagnostic error, defined as missed, wrong, or delayed diagnoses or those not communicated to patients, is common, affecting 5-10 % of hospital admissions and clinic visits. Such errors cause patient harm in up to 1 in 100 of such encounters and account for 10 % of all hospital deaths and serious adverse events. About 80 % of diagnostic errors are potentially preventable, most resulting from flaws in clinician reasoning in formulating and testing diagnostic hypotheses. The advent of artificial intelligence (AI),…
Hospital Militar Doctor Ramon de Lara, F.A.D by Cheposo. CC BY-SA 3.0 · https://creativecommons.org/licenses/by-sa/3.0
A July 2026 narrative review in Diagnosis examined whether artificial intelligence and large language models can reduce diagnostic error in bedside and clinic consultations. It summarized evidence that diagnostic errors affect 5-10% of admissions and visits and contribute to patient harm and hospital mortality.
The authors concluded that AI tools have matured enough to improve clinician decision-making and help institutions increase diagnostic safety, while emphasizing that LLMs should complement rather than replace clinician reasoning and that evolving models require continuous monitoring.
- Diagnostic error affects 5-10% of hospital admissions and clinic visits and accounts for 10% of hospital deaths and serious adverse events.
- About 80% of diagnostic errors are potentially preventable and stem from flaws in clinician reasoning in formulating and testing hypotheses.
- Review examined AI and LLMs specifically within clinician-patient encounters at bedside or clinic.
AI and large language models have matured to improve clinicians' diagnostic decision-making during bedside and clinic consultations and help institutions increase diagnostic safety, addressing preventable diagnostic errors.
The rundown
The review frames diagnostic error as missed, wrong, delayed, or uncommunicated diagnoses, common in hospital and clinic settings, causing harm in up to 1 in 100 encounters. It attributes most preventable errors to clinician reasoning flaws.
It evaluates contemporary state-of-the-art research on AI and LLMs to answer seven clinician-relevant questions about adoption, concluding LLMs should complement rather than replace nuanced clinician reasoning.
Sources
- Peer-reviewedDiagnosis2026-07-22
How should this claim be treated?
The debate