This page was generated by AI. It is not a source of trustworthy medical information.
The curated page (written by a real person) is here
other ai pages: dizzy_epi | index | outline

This comparison examines the curated page at dizziness-and-balance.com/disorders/emergency.html (2744 words) and the AI page at dizziness-and-balance.com/ai/disorders/emergency.htm (522 words). The AI page carries a byline stating it was written by Gemini AI.

Content quality

The curated page is a dense, evidence-based clinical reference. It reports specific epidemiological figures, including that dizziness accounts for 3.5% to 11% of emergency department visits, that otologic/vestibular causes make up about one-third of ED dizziness cases, and that stroke accounts for roughly 4% of dizziness presentations. It reviews triage categories proposed by Kerber (2009), discusses diagnostic accuracy and misdiagnosis rates, evaluates the HINTS and ABCD(2) decision tools, documents poor ED performance in diagnosing BPPV, and analyzes the costs and diagnostic yield of CT and MRI. It also adds critical commentary, noting for example that the steroid burst for vestibular neuritis "has now dropped out of favor" and that typical ED CT scans are "quick/dirty."

The AI page is a brief general overview. It defines common symptom descriptors, describes an "otoneurological approach," and concludes that a thorough history and targeted physical examination are the cornerstone of ED management. It contains no epidemiological figures, no named studies, no discussion of misdiagnosis rates, no decision tools, and no imaging-yield analysis. Its tone is generic and tutorial rather than clinical.

Journal article references

The curated page contains 27 journal article entries in its reference list, all from peer-reviewed publications including Stroke, Mayo Clinic Proceedings, Annals of Emergency Medicine, and The Lancet. A few in-text citations do not have matching entries in that list, such as Stanton et al. (2007) and Saber et al. (2011), and the list spells "Hansen" where the body text says "Hanson." The AI page contains no references to journal articles at all.

Authority

The curated page is attributed to Timothy C. Hain, MD, states a page modification date, and grounds its claims in 27 journal references and named quantitative sources. It is written for a clinical audience, takes positions on contested issues, and cites specific error rates and imaging yields. The AI page is attributed only to Gemini AI, presents no clinical credentials, cites no literature, and closes with the statement "This article provides general information and is not medical advice." On the available text, the curated page is substantially more authoritative.

Hallucinations in the AI page

Judged only against the text of the curated page, the AI page contains several claims that are contradicted by, or entirely unsupported in, the curated page's content:

• "Millions of visits annually." The curated page's largest reported sample is 9472 dizziness cases over 13 years; it reports proportions of visits (3.5% to 11%) and dollar costs, but never a figure of millions of annual visits.
• Symptom-quality distinction as "the crucial first step." The curated page cites Newman-Toker (2007) and Stanton et al. (2007) to argue that emphasizing symptom quality is associated with high-risk reasoning that may predispose to misdiagnosis.
• Vestibular migraine as a "common peripheral cause" managed in the ED. The curated page mentions migraine only as the main cause seen in neurology clinics, does not classify it as peripheral, and does not include it in the ED causal breakdown.
• BPPV "can often be diagnosed and treated" with repositioning maneuvers in the ED. The curated page reports that a Dix-Hallpike test was documented in only 3.9% of dizziness visits and a canalith repositioning maneuver in 0.2%, and states that ED doctors "don't seem to be very good about this."
• "Advanced imaging like MRI can be definitive." The curated page reports that the HINTS bedside examination is more sensitive than early MRI diffusion-weighted imaging, and that MRI changes the diagnosis only up to 16% of the time.
• "Multiple sclerosis exacerbation." Multiple sclerosis is not mentioned anywhere in the curated page's text.

This page was generated by AI. It is not a source of trustworthy medical information.
The curated page
see also: dizzy_epi | index | outline