Googling Medical Topics Returns Decent Results
Why this work is in the frame
A frame that forgets how it found something cannot be audited. These are the routes that admitted this work.
Bibliographic record
Abstract
FigureIn the 10 years that this column has appeared in Emergency Medicine News, we've made a point of following the development of the Internet, and promoted specific elements that can aid emergency physicians, with a special emphasis on immediate use at the point of care. We wouldn't be surprised to hear that many readers think Google is better. “I just Google whatever topic I'm looking for,” many say. But what do you get when you Google medical topics? Mostly, Google returns links aimed at laypeople, covering unrelated synonyms, pornography sites, and some direct hits. How does it compare with what's available for experienced Internet users who know how to use sophisticated search technology? What about some of the better medicine sites? We've taken four topics — Lyme disease, vertigo, post-lumbar puncture headaches, and methanol poisoning — to see what a simple Google search returns and what is available on subscription sites, through other more focused search engines, and in archives. How do you get where you want to go quickly? One of the ways Internet search engines find web pages is by indexing them according to keywords. HTML coding includes a meta tag, which is where keywords are entered by the page creator. This portion of the head section of a web page is invisible to us but not to the search engine. Using more than one keyword in your search will modify the return you receive. Recommendations for Lyme disease continue to evolve. Our initial Google search on Lyme disease brought thousands of links. One of the first took us to the Centers for Disease Control and Prevention, which has many useful professional links. Another good link from the American College of Physicians for ACP Online (www.acponline.org), appeared as the 23rd link. The eMedicine Lyme disease article appeared as the 38th returned link. (http://emedicine.com.) Using additional search terms for the Lyme disease search had mixed results. Adding the word “professional” returned a decent piece on best practice from the online Merck Manual but little else. Using the terms “therapeutic,” “review,” “research,” and “diagnostic” in the simple Google format did not substantially improve the links returned. The returns did improve, however, when we used some of the more sophisticated search parameters available on Google. Simply restricting the searches to a recent time frame, from 2003 to 2006, improved the quality of returns in the previous samples. The CDC site provided by Google on the first search is probably the best bet. The Cochrane Collaboration (www.cochrane.org) is a good source of evidence-based recommendations, if the site reviewed the topic you seek. In the case of Lyme disease, Cochrane only addresses use of the vaccine. MDConsult (www.mdconsult.com), a subscription service, will take you fairly quickly to the Infectious Diseases Society of America's guidelines. Once you know what you want, you can often find it with Google. For example, the same document is available for free if you search for “Lyme IDSA guidelines.” Core EM Topics Methanol poisoning is a core topic in emergency medicine. We have to know about it for the board exam, but beyond that we may not deal with it very often. When faced with a patient who ingested methanol at 2 a.m., we might like to see a quick review to be sure that we haven't forgotten some crucial aspect of management. Of course, the universal number for poison control in the United States is worth a plug: 1-(800)222-1222, and some centers provide excellent advice while others are staffed by nonclinical drones reading from a script. Searching for “methanol poisoning” on Google provides as the sixth return a link to an excellent review from the Canadian Journal of Emergency Medicine and as the seventh return a link to a web site for the relatively new pharmaceutical antidote fomepizole (http://antizol.info/mpoisono.htm). In this case, our preference is to read the chapter in Rosen's available in full text online through MDConsult or through a separate subscription to www.rosensemergencymedicine.com. With Rosen's, you know what you are getting. Similarly, Tintinalli's slightly more concise text is also online through McGraw Hill's Access Medicine (www.accessmedicine.com) for $150 per year. McGraw Hill offers a host of other excellent texts, but not yet Goldfrank's Toxicologic Emergencies. Perhaps with the next edition. Looking up “post-lumbar puncture headache” is a good illustration of the various approaches to medical searches online. An exhaustive literature search of the existing evidence is available through PubMed.org. Be prepared to wade through several hundred titles, and then take a trip to the library. The shortest and least expensive route to information is with Google. The search term “lumbar puncture headache” produces several excellent links, including an evidence-based review in www.pulmonaryreviews.com that debunks the recommendation to have patients lie flat after the procedure. (It's the needle, stupid!) There also is a historical essay from emedicine.com. The challenge to using these quick and easy links is that you cannot easily evaluate their quality. On the other hand, if you accept the Cochrane method, you will trust their abstract on the subject of bed rest. They found no evidence of benefit in 15 randomized controlled trials involving 1001 patients. For medical information searches, you will find some of the best results through the Google scholar option (http://scholar.google.com). From that search page, choose the advanced scholar search (http://scholar.google.com/advanced_scholar_search?hl=en&lr=) to find the best returns. Why? You'll notice that you can restrict your search to Medicine, Pharmacology, and Veterinary Science. Returns from this section include abstracts or full-text articles from multiple journals, including the New England Journal of Medicine, the Annals of Internal Medicine, the Journal of the American Medical Association, and Science. In most instances, you will have to pay an exorbitant fee for immediate access to the full-text articles. Pay the $20 or return to the web, and keep looking. You may well find it for free. Searching for the term “vertigo” represents a challenge to any search engine. Where to start? A simple search for “vertigo review” will produce a lot of links about the Hitchcock movie, Vertigo. “Vertigo, research” works well, as does “vertigo therapeutics,” which took us immediately to the eMedicine article. We repeated the process with the scholar search page, restricting by years and using the Medicine, Pharmacology, and Veterinary Science category, and had much better success with “vertigo, review.” For those readers who have set up their own home page, the scholar search would be a great page to add. Please note two other features of the Google Scholar Search. The number of times the article is cited by other works in the database is listed below each result. Clicking on the “cited by x” link takes you to all the links that refer to the work. In the search for “vertigo,” you will arrive at a link regarding the Epley maneuver. Using this feature works as a double search, isolating links that are broadly interesting to you regarding the topic of the first link. In this case, you receive all the articles citing the article about the Epley maneuver. Clicking on the “web search” link next to the “cited by” link will take you to all the sites directly related to the Epley maneuver itself, branching into another search if you are interested without changing the search parameters. Scholar Search is currently in beta mode, and it will be interesting to see how it develops. You'll also notice an option called Images above the simple Google search bar. Clicking on this after entering Lyme disease returns an assortment of images of erythema chronicum migrans and the Ioxodes scapularis tick. This is another very useful feature of Google.FigureFor those of you who can tolerate no more than one page of results, you can increase the number of returns per page from 30 to 50 in the Google preferences page (www.google.com/preferences?hl=en). These preferences can be saved in a cookie on your browser. The display font size doesn't change, and you won't spend so much time moving from page to page. Just a few years ago, looking for answers to medical questions required a trip to the library. Not so in 2006. Medical content on the Internet has proliferated to such an extent that Google can find answers to many of our questions. Along with links to sites that address your questions, you will have to wade through advertisements and patient-oriented informational web sites, among other things. Several subscription sites will provide more refined results, including some of the best textbooks available in full text. The scope of medical literature can be surveyed with PubMed.org from home or in the ED.
Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.
Full frame distilled prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. Learned from the 10,348 direct Codex labels and 10,348 direct Gemma labels. Candidate is the union of thresholded teacher heads; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels or direct frontier model labels.
Codex and Gemma teacher scores by category
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.001 | 0.003 |
| Meta-epidemiology (narrow) | 0.000 | 0.000 |
| Meta-epidemiology (broad) | 0.001 | 0.000 |
| Bibliometrics | 0.000 | 0.001 |
| Science and technology studies | 0.000 | 0.000 |
| Scholarly communication | 0.000 | 0.000 |
| Open science | 0.000 | 0.000 |
| Research integrity | 0.000 | 0.000 |
| Insufficient payload (model declined to judge) | 0.011 | 0.000 |
Machine scores (provisional)
The two teacher heads of the student model, read on this work. A score orders the frame for review; it never asserts a category, and the validation status ships verbatim with every row.
Baseline scores from an immature model (maturity gate not passed, 7 training rounds). Scores rank; they never assert a category.
score_only:v0-immature-baseline · verbatim from the scoring run: score_only means the number may rank works, and no category label ships from it