Tammy Ross

Welcome everyone and thank you for joining today's webinar From Clinical Question to Confidence Search, Teaching Evidence Retrieval with MEDLINE Ultimate. I'm Tammy Ross, Senior Product Marketing Manager at EBSCO. Today's presenter is Kathryn Vela, EBSCO Senior Customer Training Specialist.

Kathryn has nine years of experience as a medical librarian in hospital, academic, corporate, and research environments.

She holds an MLIS degree from San Jose State University and a BS in biology from Washington State University, as well as a System Systematic Review Services specialization from the Medical Library Association. So without further delay, I will now turn the presentation over to Kathryn.

Kathryn Vela

Hi, everyone. I'm excited for today's session. We're going to be talking about some very cool tools and features from MEDLINE Ultimate.

We're going to cover how to use the PICO search tool, the mesh vocabulary, AI insights, and natural language search mode to create effective search strategies in MEDLINE Ultimate.

In practice, I often introduce these approaches in stages, starting with natural language for accessibility, then moving into structure frameworks like PICO, and finally into more advanced strategies like mesh and Boolean searching as students build confidence.

We're going to start today with the PICO search tool. This is a tool that helps users to create focused, searchable clinical questions using guided search boxes that follow the PICO framework.

And if you're not familiar, the PICO is a framework that helps researchers to break down their research questions into its most important component parts. So that's your population, problem or patient, the intervention of interest, a comparison intervention, the outcome or outcomes of interest, and then a time frame or perhaps the type of study that you're interested in.

This is a framework that many educators use to help students move from a broad or unclear clinical question to something that's structured and searchable. And it's especially helpful for learners who are just starting to engage with evidence-based medicine.

This screenshot shows you what the user will see in EBSCOhost, or EDS, EBSCO Discovery Service, if you have that platform as well, when they use the PICO guided search tool. All they need to do is enter their search concepts into the appropriate box, and when they're ready, they can select add to search.

When using this form, just note that the p and the I fields are required. Everything else is optional.

Another handy feature to know about is the MeSH terms in MEDLINE. These are the National Library of Medicine's controlled vocabulary for the MEDLINE database, and it includes over nineteen thousand main headings that are arranged into a hierarchical tree structure.

Using MeSH terms in your search can help to improve the precision and relevance of your search results.

From a teaching perspective, this is also a great way to help students understand how medical concepts are related.

Exploring the hierarchy can surface broader or narrower terms, which often leads to better research questions.

To access MeSH terms, all you need to do is look for the MEDLINE MeSH twenty-six heading on the advanced search page. You'll then have the option to execute keyword searches for the terms you're looking for and take deeper dives into those terms to learn more about how to use them effectively.

When you select a term, you're able to see the definition for that term, any additional entry terms, as well as seeing where that term falls in the term hierarchy.

The last two features we'll talk about are somewhat related.

First is the generative AI insights tool. This generates concise summaries of full text articles in two to five bullet points and pulls this information directly from the full text that is available within EBSCOhost.

This tool helps researchers quickly review the content of a research article and make a quicker decision about whether it's relevant to their research.

And this is particularly useful when teaching students how to evaluate sources efficiently. It gives them a quick way to assess relevance before committing to a full read, which is an important skill in evidence-based practice.

It's also very handy if a record doesn't have an abstract or if the abstract doesn't include a lot of detail.

And similarly, we have natural language search mode. And just as a note, in July, this tool will be renamed as AI assisted search.

So this tool allows users to enter their searches in a more conversational tone. It uses natural language processing to interpret the intent and context of these search queries.

For many students, this lowers the barrier to getting started with their search. Instead of worrying about syntax, they can begin with how they naturally ask a question and then refine from there.

And users can use both natural language search mode and a more traditional Boolean style of searching within a single session.

These screenshots show how an example show an example of how a user might leverage natural language search mode to run their search.

So instead of using keywords and Boolean operators to create their search, they can simply type out a sentence that matches their search question.

Just want to make sure that you're toggled into that natural language search mode. And then when they're looking at results, they'll also be shown a list of AI-generated suggested search queries that may also help to fulfill their research need.

So let's take a quick look at these features in action. So what I'd like to show here, is not just different ways to search, but also how you might guide students from an initial question to a more refined and effective search strategy over time.

So let's say we're interested in investigating some of the latest research on heat related illness.

We can try building a search on this topic using some of the different tools and features that we just talked about.

And we can start with the PICO tool.

So here on the advanced search page that we're currently on, we can see the PICO tool showing up here at the top right of the search boxes.

If we were in the basic search view, we could also see that here at the top right of the screen or of the search boxes.

So if I click on this, I'm now shown that guided search screen that has each element of the PICO framework with its own keyword search bar.

So I can start typing in keywords related to my research question, starting with heat related illness as our problem or condition. Maybe we're interested in hydration as an intervention, and we want to compare that to clothing as protection against heat related illness.

You'll notice I did leave some fields blank here. I'm not going to fill an outcome or time or type of study here.

In many cases, adding the outcome in the time frame can unnecessarily limit your search result, and you might miss on out on some relevant stuff or you might get too narrow of a search set.

And this is also a great time to remind folks that searching is iterative, and it may take a few tries to get to the most effective search strategy.

So let's add to search.

And we can see that we've been dropped back into the advanced search view, and the terms that we had entered into each keyword search box have been combined with and.

From here, I can continue to tweak my search by adding filters, additional terms. I could look at those MeSH terms, or I can go ahead and hit search and see what kind of results I get.

Another way to construct this search is using the natural language search mode. So let's see how that works.

I'm going to go to the basic search. This is the easiest way to get to the natural language search mode. You can see here below the search box, there's this toggle for natural language search. And when we select this, it's, very clearly turned on, so it's easy to see that you're in this search mode currently.

And then I can just type in to the search box or copy in my question. What are effective ways to treat and prevent heat related illness?

I'll hit the magnifying glass or I can hit enter to run this search, and we can see that we're still in the natural language search mode, so that toggle is still turned on.

We have this how does it work button here. If I click this, it gives me a very high-level overview of what is happening with the natural language search mode, if I was curious about that. And then show refined query shows me how the database interpreted my natural language query into a Boolean query.

And this is what was actually executed by the database. So I can check this just to make sure that it did interpret my question correctly. This is to help with transparency with this tool.

And then below, the search box area and just above the results, we have this, the AI generated suggested queries. So I can look at these and evaluate if they might be a better way to phrase my question or if they might address a part of my question that I'm actually more interested in than what I initially thought.

And if I did want to explore one of the suggested queries, I just need to click on it, and it'll run that search for me right away, still in the natural language search mode.

And to compare, if we try to run the same type of search phrasing outside of the natural language search mode, we will not get the same set of useful results. For example, we can see here we have twenty-nine results. If I just turn this toggle off for natural language search mode, it doesn't work.

So that's just why it's important to make sure that if you are going to phrase your question this way, make sure you have the natural language search mode turned on.

So starting a new search here, we're going to look at the third way to construct our search here using the more traditional Boolean style of searching.

So let's take a look how that might work if we were to incorporate the MeSH terms that we were discussing earlier.

So here below the search boxes, we can see that MEDLINE MeSH 2026.

If I click on this, now I have my keyword search bar where I can look for MeSH terms by keyword.

So I'll try heat related illness, And I hit enter to run that search, and I'll scroll down to see what kind of results came up from my search.

And I think this third option is, the closest to what I'm looking for. But just to check, I can click on it, and I can see the definition here.

Talking about conditions that developed due to overexposure or overexertion in excessive heat.

I can see some synonyms or entry terms listed here as well.

These all match what I'm thinking of for heat related illness, so this looks like the exact term I need.

If I wanted to look a little closer, I could also scroll down and look at the term hierarchy. So I can see that this term falls under wounds and injuries.

And if I keep scrolling down, here's heat stress disorders, which also, encompasses heat exhaustion, heat stroke, and hyperthermia.

So I'm going to add this term to my search by just, checking the box here. I could also check the box next to it in the term hierarchy.

And I'm going to hit add to search.

And now we can see it's been dropped into my search box here at the top. And we can also see that it's been exploded.

So that's what the plus sign here indicates, and that means that those narrower terms we just saw are also included in this term search.

Now I want to add some terms related to hydration. I'm still interested in in heat related illness and hydration as a way to prevent them. So I'm going to come here and start typing in hydration. You may get some suggested terms here, and you can just click those to add them into the search box.

So I'll just leave this at as hydration, and I'm going to go ahead and hit search.

And now we can start looking at our results.

So we have a little less than three hundred results, and I can start looking at what we have here and see if any of them match my research interest.

And one way I can improve the efficiency of this process is by leveraging that generative AI insights tool.

So let's take a look at how that looks as we're scrolling through our results list. You're not going to see it on every result, but when you do see it here, we can see it on this option here because the full text is available in EBSCOhost. On this first result, we don't see it because we have to go outside of EBSCOhost to get to the full text.

We will come down to this option here, and I'll just click on AI insights, and it starts to immediately generate that bullet point summary for me.

Together, these approaches mirror how students develop the search skills.

From asking a question in their own words to structuring it to refining it for more precision, these tools support that progression at each stage.

Whether you're working with early learners or with more advanced students, these features can be incorporated into assignments or live instruction to help build confidence and evidence retrieval.

That's everything I wanted to cover with you today. I'm going to pass it back over to Tammy before we open up to questions.

Tammy Ross

I'm going to open it up to questions, and I did see a few already come through. So Kathy asks, how are AI insights different or better than abstracts?

Kathryn Vela

That's a great question. I wouldn't say they're better, but they are different.

Abstracts are going to be submitted by the author. They're included with the publication, but there can be a lot of variation with the abstracts. You know, ideally, you would get a structured abstract that includes a lot of detail about the, you know, research question and the methodology and the results.

But other times, you get abstracts that are more thin. Maybe they just say what the research question was, but they don't share anything about the methodology or the results. And it can be hard to evaluate that study based on very limited information from the abstract in that way. So, the AI insights can help to fill that gap. That's my favorite use for them when, you know, the abstract isn't really giving me what I need. I can look at the AI insights, and that will tell me more about what's going on in the study without me having to dive into the full text and parse that out for myself.

If there is a really good structured abstract, you can still use the AI insights just as an additional tool to flesh out the study more fully before deciding if you want to dive into the full text.

Tammy Ross

Awesome. And Thomas asked if this feature is also available on CINAHL, and I believe it is.

Kathryn Vela

Yes.

Tammy Ross

Awesome. Shannon asks, how do you get PICO or natural language search turned on for your institution?

Kathryn Vela

So that is just a really easy setting to turn on in EBSCO Experience Manager, and I can drop that in the chat. Is it okay if I do it in the chat, Tammy?

Tammy Ross

Yes. Yeah. Make sure you've got to everyone selected so that they all see it.

Kathryn Vela

Got it.

Tammy Ross

Host and panelists is defaulted. Thank you.

Kathryn Vela

Yeah. So I just dropped in the chat a link to some EBSCO Connect documentation. It covers how to configure multiple search settings in EBSCO Experience Manager, but that does include PICO.

Tammy Ross

Awesome. Okay. Here's another question. What should students do if they don't get enough results?

Kathryn Vela

Really good question. So that happens a lot. It's a totally normal part of the searching process. When I work with students, I know that's something that tends to freak people out and they start to panic a little bit.

So that's where I just really like to emphasize that this is a normal part of the searching process. It's very experimental. I mentioned earlier that it's iterative. So in the case that you're not getting enough results, that's your clue to start broadening your search structure, your search query.

Start taking some terms off. If you included terms related to time frame or type of study or outcome, maybe take those off and try running your search again and see if that brings back more stuff for you. If you have a comparator in addition to an intervention, maybe try oring them together instead of and, and that might bring in some more stuff for you.

Maybe you try everything you could think of, and you're still not getting enough results.

It's not super helpful in the moment, but that is an indicator that, you know, that's a gap in the research.

And, you know, maybe more attention needs to be paid to that research question. And you, in the moment, need to think about maybe broadening or switching gears with your question.

Tammy Ross

Awesome. And I just saw Trace pop up in the chat. PICO and AI Insights, are they available in MEDLINE Complete?

I think so because it's the it's the interface, not the database in particular.

Kathryn Vela

I believe so. Yeah.

Tammy Ross

So how would you recommend introducing MEDLINE Ultimate to first year students who are completely new to database searching?

Kathryn Vela

Yeah. The I think the key for new students is taking it slow and not introducing too many new concepts or tools at once, which can be easy to do, especially if, you know, you as the expert are trying to share the information. It can seem like a very easy thing share with somebody else. But if they're not coming to the same foundation that you are, it can be a lot. It can be overwhelming. So I like to go with, shorter sessions if I can, if I'm doing, like, a workshop, maybe just a thirty-minute workshop.

If I'm able to put together video tutorials, keeping those very short, three to five minutes tops.

And then also offering perhaps text-based guides as well. So offering multiple modes of learning too so you're able to really meet the students where they're at. And then just kind of building up that scaffold of, you know, starting with constructing a basic keyword search and then talking about how to apply filters and then talking about how to add those MeSH terms, all in, like, different sessions, different videos, different tutorials so that they can start to build on their own learning.

Tammy Ross

Awesome. And do you typically suggest starting with natural language search or jumping right into structured approaches like PICO?

Kathryn Vela

That, I think, depends on what your goals are with your research.

If you are just doing, like, a quick exploratory search, then the natural language search mode can be a great place to start. It's quick. It's easy. And those suggested queries can help you to more fully wrap your head around what exactly you want to research. It's kind of like walking yourself through a reference interview almost.

And so we can help to really narrow down what it is you're looking for.

But if you know you're doing something a little more structured, like a literature review, a systematic review, scoping review, anything like that, then I would want to start with a more structured Boolean search. But even in that case, if it's a new topic for you, starting with the natural language search mode to develop more of a familiarity with the literature and what's available, and then switching to the Boolean searching is also a really good option.

Tammy Ross

Karen asked if you could just show us one more time where the AI insights is found on MEDLINE. Yeah. Absolutely. Can the screen.

I'll stop my share.

Kathryn Vela

Yeah. So you'll see it on the results page.

And if you're not seeing it, that's also a setting you need to turn you might need to turn on, And it's also an EBSCO experience manager.

So, again, you're not going to see it on results that don't have full text within EBSCOhost. This tool is staying, like, within the confines of the EBSCOhost database. It's not going to go out into, like, full text finder or anywhere else you have full text. So in this first example, the full text is out in full text finder, so you're not going to see the AI insights button show up here.

But you do see it here with the second option because the full text is within EBSCOhost, and then the button disappears. And you click on it, and it will generate, that bullet point summary for you. And if we scroll down a little more, we can see sometimes we'll see it and sometimes we won't. It just depends on where the full text lives.

Tammy Ross

Awesome. You're welcome, Karen. Can users switch between natural language and Boolean searching within the same session?

Kathryn Vela

Yes. Absolutely. You can do that. It's just a matter of toggling that natural language search mode on and off. Just know that if you, like, start in the natural language search mode and then you decide to switch out of it, you'll also want to switch the way you've constructed your search. You don't want to do that natural language phrasing anymore. You want to do your Boolean ands and ors.

And you can use that refined query, that show refined query button that I showed you that displays the Boolean query that the database sets up based on your natural language question. You can copy that, put it into the search box, switch out of the natural language search mode, and then you have a built in ready to go Boolean search that you can work with, tweak, edit, however you want to use that.

Tammy Ross

And I think we have time for maybe one more question. When should students move from natural language searching to using MeSH terms?

Kathryn Vela

That's an interesting question. Part of it is preference, and part of it is, like I said earlier, the goal of what you're searching for. It may also depend on the assignment. If they're doing a search for a particular assignment, maybe there are requirements about which or how they do their search. So that's a factor as well.

But I think what I was saying earlier also applies, you know, if they're just doing kind of more exploratory searching, natural language search mode is a really quick and easy way to get going on that. But if they do want to do something more structured for a more structured type of assignment, then switching over to the Boolean searching would probably be a better option.

Tammy Ross

Great. Well, if there I don't see any other questions anyway, so our timing is great. It's two twenty-eight, and we only had two minutes left. So I will just take this opportunity to thank everyone for joining us for this sort of bite sized micro webinar today. I have included our email addresses on the slide this last slide. So if you have any questions that come up after we've all gone about our days, you're welcome to email, Kathryn or myself directly. So thank you again, and you will receive a recording of this, later this week. And thank you again for attending, and I hope you have a great rest of your day. Thank you.