Research Mode: what the evidence says

What Research Mode does

Research Mode turns an evidence question into a cited summary of what published research and guidelines say about it. On the Home screen you switch from Ask Anything to Research, and the screen changes to "Research the literature: find studies, guidelines, and evidence with in-depth source analysis." While the answer is written, the steps shown are Finding research, Ranking study quality and Summarizing evidence. The answer names the key trials, meta-analyses and guideline positions, gives their populations and results where the sources state them, and says where the evidence is weak or conflicting and which evidence is stronger. Every claim carries a numbered citation, as in any AskMed answer. Ask Anything is better for concepts, mechanisms, differentials and drug facts; Research is for questions such as whether a treatment works, how two approaches compare, or how strong the evidence behind a recommendation is. Both modes write between 150 and 1,000 words, as the question needs.

• Good Research questions: "Does antibiotic timing matter in sepsis?", "DOACs vs warfarin in atrial fibrillation?", "Is colchicine useful after a heart attack?"
• Method questions work too: "What makes a meta-analysis reliable?" or "How is bias assessed in systematic reviews?"

How Research Mode works

Research Mode uses the same search as Ask Anything and then reads further. AskMed writes a precise search query, adding words such as "review" or "guideline" for evidence questions, and searches its fixed list of medical sites, including PubMed and PubMed Central, the Cochrane Library, major journals, guideline bodies and drug labels. In Research Mode, AskMed then opens the most relevant result and reads its full text, not only the short excerpt the search returns, before the answer is written. The AI model is told to prefer guidelines and health agencies, then systematic reviews and meta-analyses, then review articles and references, then large trials and other studies, and to prefer work from the last 10 years unless the topic is settled. When guidance has changed, the answer gives the newest position, says what changed and since which year, and names any society that still disagrees. You choose how many sources it cites, from 3 to 12.

The study type filters matter most here. Turn off Other to leave out observational studies and case reports, or keep only Meta-analysis to see what systematic reviews concluded. The source filters page explains each switch.

When Research Mode helps

Research Mode helps when the question is about evidence rather than facts. A physician preparing for journal club can get the landmark trials and meta-analyses on a topic in one cited list, then open each paper. A resident asked on rounds why a guideline changed can see what the newer guidance says and since which year. A pharmacist can check what trials found on a drug class for a given condition before a formulary discussion, without any dose figures. A nurse writing an evidence-based practice project can find the reviews behind a nursing intervention and cite them properly. A medical student can learn how to judge a study by asking how bias is assessed or what makes a meta-analysis reliable. A researcher scoping a new question can see quickly where the evidence agrees, where it conflicts and where it is thin, and decide what to read first in full.

For a comparison with another AI search app's research features, see AskMed vs MediSearch. Get AskMed on the App Store for iPhone and iPad, on Google Play for Android, or use it on the web.

Limits of Research Mode

Research Mode is a fast, cited summary, not a systematic review. AskMed runs one search per question and reads one source in full; the other sources are used through the excerpts the search returns, so details deep inside those papers can be missed. The search covers a fixed list of medical sites, so studies found only in other journals or databases may not appear, and AskMed does not grade evidence with a formal tool such as GRADE. The AI model writes the summary and can misread a result or overstate it, and no person reviews the answer before you see it. Study types are judged from each article's title and excerpt, so a filter can occasionally let the wrong kind of study through. Research Mode gives no drug doses, even when a trial reports them. For anything you will act on, publish or teach, open the cited papers and read their methods and results yourself.