The research agenda has helped shape what evidence gets produced.
But clinical questions reveal something different: the demand side of evidence.
What do clinicians actually need to know when evidence meets real-world care?
https://t.co/UWkjcMwLwe
If used carefully, clinical Q logs could help researchers, guideline developers and medical AI teams.
But they are not simple truth, they are signals.
They reflect who uses the system, what the interface encourages, and what people feel able to ask.
https://t.co/UWkjcMwLwe
Clinical questions are signals of friction.
Sometimes the evidence is missing.
Sometimes it exists but is hard to find, hard to interpret, hard to trust, or hard to apply to the patient in front of you.
We need to understand that demand side better.
https://t.co/UWkjcMwLwe
AskTrip has now received ~19,000 clinical questions.
What stands out is not just the range of topics, but the range of reasons clinicians ask.
Some questions are about missing evidence. Many are about applying existing evidence to messy real-world care.
https://t.co/UWkjcMxjlM
The production side of evidence is relatively well organised: trials, reviews, guidelines, summaries.
The demand side is much less visible.
Natural-language clinical questions may help us understand where clinicians are uncertain in real practice.
https://t.co/UWkjcMxjlM
Evidence-based medicine often asks: what evidence exists?
But perhaps we should also ask:
What are clinicians repeatedly trying to find out?
That may tell us where evidence is failing to connect with practice.
https://t.co/UWkjcMwLwe
Search logs tell us something.
“Atrial fibrillation elderly” gives us a fragment.
But full clinical questions reveal more: the uncertainty, the context, and what the clinician is really trying to find out.
https://t.co/UWkjcMxjlM
A repeated clinical question is not automatically a research gap.
It may be a dissemination gap, an applicability gap, a guideline ambiguity, or a problem translating evidence into action.
That distinction matters.
https://t.co/UWkjcMwLwe
Much of evidence production is shaped by the research agenda.
But clinical practice creates its own agenda too.
It shows up in the questions clinicians repeatedly ask.
That demand side deserves attention.
https://t.co/UWkjcMxjlM
Evidence-based medicine has spent decades improving the supply side: trials, systematic reviews, guidelines, summaries and search tools.
But what about the demand side?
What are clinicians actually trying to find out when they reach for evidence?
https://t.co/UWkjcMxjlM
Search alone isn’t simple. Keyword precision, semantic understanding, source authority and recency all matter. Our latest blog explains how we’re testing a hybrid approach on Trip.
https://t.co/0aHy6fYgaJ
We’re not rushing this into live use. First, we’re expanding offline testing across different query types so we understand where each search method works — and where it fails.
https://t.co/0aHy6fXIlb
Search in clinical evidence is tricky because the same concept can be described many ways: heart attack, myocardial infarction, MI.
Hybrid search may help bridge that gap.
https://t.co/0aHy6fXIlb
“More semantic recall” sounds good. In our early testing, it mostly added noise. Better search is not always about casting a wider net.
https://t.co/0aHy6fYgaJ
The future of search on Trip probably isn’t one universal ranking method. It’s likely to be smarter routing: lexical-led for simple broad queries, hybrid + boost for richer ones.
https://t.co/0aHy6fYgaJ
Hybrid search performed well for complex clinical queries. But for broad single-term searches like “asthma”, plain lexical search did better. That was the interesting bit.
https://t.co/0aHy6fYgaJ
AskTrip’s latest upgrade includes Explore further, improved evidence scoring, fuller default answers, and PDF downloads.
Phase Three testing is now underway.
https://t.co/pnF06M8oJR
One early lesson from our search testing: boosting for authority and recency really matters. In evidence-based medicine, the best current evidence should rise to the top.
https://t.co/0aHy6fYgaJ