Evidence companion
inScope
A more concise alternative to guideline reference tools like UpToDate, inScope provides Quick Answers for narrow questions or evidence-based Mini-guides for inpatient management decisions.
This tool is an AI-powered RAG system that synthesizes high-quality clinical evidence. It is an aid and does not replace clinical judgment. Always integrate findings with patient-specific factors and local practice standards.
Tip: Start with a short, specific question (e.g. Antibiotic choices for inpatient CAP in a 65-year-old with penicillin allergy?).
Created and maintained by Yasmine Abbey, MD, MSc · Last reviewed:
Educational decision-support for clinicians. Not medical advice — verify all outputs against your clinical judgment, institutional protocols, and current guidelines.
How inScope answers a clinical question
inScope is a free, physician-built evidence lookup for inpatient medicine. You ask a focused clinical question — 'antibiotic choices for inpatient CAP in a 65-year-old with a penicillin allergy', 'when does hyponatremia need hypertonic saline' — and it returns a short, structured answer with the sources it used: a Quick Answer for a narrow question, or a mini-brief with a bottom line, the evidence, caveats, and citations for a bigger management decision.
It is designed as a concise alternative to reference tools like UpToDate for the moment on rounds when you need a defensible answer in minutes, not a chapter. It is not a replacement for those references, for the primary literature, or for your own judgment, and every brief says so.
What happens between your question and the brief
A brief is built in stages rather than by a single model call. First the question is translated into a structured search: the clinical problem, the population, the decision at stake, and the terms a guideline would use. Then the system retrieves candidate sources — society guidelines first, then systematic reviews and primary studies from PubMed and Europe PMC, preferring open-access full text and reading the documents it cites rather than summarizing abstracts. An excerpting step pulls the passages that bear on the question. A reasoning step writes the answer from those passages only, and a separate verification step checks each claim against the quoted evidence and grades the result.
That last step is why a brief carries a verdict. A SUPPORTED brief means the central recommendations are backed by quoted guideline or trial text; PARTIAL means some claims rest on weaker or indirect evidence; INSUFFICIENT means the sources retrieved could not settle the question, and the brief says what is missing rather than guessing. The clinical assessment and the evidence verification are reported separately so you can see where the reasoning is strong and where it is thin.
Which sources it reads, and in what order
- Society practice guidelines and their derived documents (ATS/IDSA, ACC/AHA, KDIGO, European Society guidelines, NICE, and others) are retrieved first and anchor the brief.
- Systematic reviews and meta-analyses, then primary trials, from PubMed and Europe PMC.
- Open-access full text is preferred; paywalled sources are resolved by DOI to a legal open version when one exists, and otherwise cited from the abstract with that limitation stated.
- Each citation is shown with its evidence tier, year, and a verbatim supporting quote, so you can judge the source without opening it.
- Sources are checked for currency: superseded editions and sibling documents from the same society are distinguished so a 2014 guideline does not stand in for its 2023 update.
Quick Answer or mini-brief: which one you get
Narrow questions with a factual answer — a dose, a threshold, a first-line agent — get a Quick Answer: a few sentences with citations, in seconds. Management questions that involve a decision get a mini-brief: a one-paragraph bottom line, an evidence summary organized around the decision, caveats, and references. When a critical branch of the decision is missing from your question (is the patient pregnant? on dialysis? already on anticoagulation?), inScope asks for it rather than answering both branches at once. Follow-up questions keep the thread, so you can narrow or extend a brief without starting over.
How inScope compares to UpToDate, OpenEvidence, and a general chatbot
UpToDate is a curated, peer-reviewed reference written by expert authors and updated continuously; it is the standard for depth and is paid. OpenEvidence is a free, retrieval-based answer engine for verified clinicians with a large licensed literature base. A general-purpose chatbot answers from memory and may cite sources that do not exist. inScope sits between them: free, no login, built around guideline-first retrieval and an independent verification step, and transparent about what it could not find. It is narrower than UpToDate, newer and smaller than OpenEvidence, and more cautious than a chatbot. The Resources section has a longer comparison of free UpToDate alternatives.
Limitations you should know before relying on a brief
- A brief is only as good as the sources retrieved. Rare topics, very recent trials, and questions outside inpatient adult medicine return thinner briefs, and the verdict will say so.
- Doses and thresholds are quoted from sources, not computed; confirm them against your formulary and the patient's renal and hepatic function.
- A brief can take one to three minutes because it reads full documents; the progress bar shows which stage it is in.
- Do not enter patient identifiers. Questions are processed to produce the brief and are not used to train any model; see the privacy policy for details.
- inScope is educational decision support for clinicians. It is not medical advice and does not replace institutional protocols or specialist consultation.
Frequently asked questions
Is inScope really free, and do I need an account?
Yes, and no. inScope is free to use without a login. It is funded by its physician creator and by voluntary support from users. Usage limits may apply at times of high demand to keep it free for everyone.
Is inScope a free alternative to UpToDate?
It is a free alternative for one specific use: getting a concise, cited answer to a focused inpatient question quickly. It does not replace UpToDate's depth, breadth, or editorial review, and it is not a replacement for reading the guideline when the decision is high-stakes.
Where do the citations come from, and can I trust them?
Citations come from retrieved documents, not from the model's memory: society guidelines, systematic reviews, and primary studies indexed in PubMed and Europe PMC. Each citation shows its evidence tier, year, and a verbatim supporting quote, and a separate verification step checks the brief's claims against those quotes. You should still open the source for any decision that depends on it.
What does the verdict (SUPPORTED, PARTIAL, INSUFFICIENT) mean?
It grades how well the quoted evidence supports the brief's central recommendations. SUPPORTED means they are backed by quoted guideline or trial text; PARTIAL means some claims rest on weaker or indirect evidence; INSUFFICIENT means the retrieved sources could not settle the question and the brief states what is missing. The clinical reasoning and the evidence check are reported separately.
How is inScope different from asking ChatGPT or another chatbot?
A general chatbot answers from training memory and can produce confident text with invented references. inScope retrieves real documents first, writes only from the passages it extracted, verifies each claim against them, and tells you when it could not find enough evidence. It is slower and more cautious by design.
Why does a brief take a few minutes?
Because it reads the documents it cites rather than summarizing abstracts. Retrieval, full-text reading, excerpting, reasoning, and verification run in sequence, with parts in parallel; a Quick Answer for a narrow question is much faster than a full mini-brief.
Can I ask follow-up questions?
Yes. Follow-ups keep the thread of the previous answer, so you can narrow the population, add a comorbidity, or ask about the next step in management. If a critical detail is missing, inScope will ask you for it before answering.
Can I use inScope with patient information?
Do not enter identifiers. Describe the clinical scenario in general terms (age range, relevant comorbidities, the decision at stake). If you need to work from a note, run it through the PHI scrubber first, which de-identifies text entirely in your browser.
How are the clinical guides on this site related to inScope?
Each guide in the Guides section was generated with inScope from a focused question, then clinically reviewed by a physician before publication. They show what a brief looks like and are kept as a reviewed reference library; inScope lets you ask your own question.
Further reading
- scopeMD Resources: What are the best free UpToDate alternatives in 2026? A comparison of inScope, OpenEvidence, Medscape, Merck Manual Professional, DynaMed, and UpToDate.
- Guyatt GH, Oxman AD, Vist GE, et al. GRADE: an emerging consensus on rating quality of evidence and strength of recommendations. BMJ. 2008;336(7650):924–926.
- scopeMD Patient Privacy & PHI De-Identification Policy.