Scite MCP Clinical Trials Demo

Hi there.

I'm Drew Barontini, VP of Product at Scite, and I want to show you a demo of how you can search for clinical trials inside of the Scite MCP.

Now, we launched the MCP across our broad range of academic literature and we've expanded that to access to other data sets like clinical trials.

Now, you can use this inside of Claude or ChatGPT or any other tool that works with MCP servers.

So I'm going to post a question here and I'm going to run this query to search for phase 3 clinical trials specifically around this GLP1 receptor and specific treatments in obesity or type 2 diabetes.

I'm asking it for a structured table, I'm asking it for specific information, and I wanted to summarize the general competitive landscape, so understanding who the main sponsors are and what interventions are being tested.

So this is going to look through our aggregate data set across clinical trials, and you can see that Claude is figuring out what it should call running its calls against the Scite database and data set, pulling that information back and then orchestrating a response.

All right, so you can see here that we pull back that information.

So it's showing the phase 3 clinical trials and showing that information in a table like we specified.

It includes the ID information, the title of the trial sponsor, phase, start date, estimated completion dates, and the intervention.

So you can obviously specify whatever you want for Claude to come back with that data and structure it however you'd like to see it.

Which is the benefit of using an MCP is you can refine the response to map to whatever you're looking to do and how you're looking to visualize that information.

So you can see all that information is listed here in this full table response, and it's talking about how it went through.

And it's also painting a picture of the competitive landscape to understand there are some, specific pharma companies shaping some of this.

It's talking about other related competitive information and providing additional context based on the questions that were asked.

Now, what's interesting here is to extend this past just the, the information around clinical trials and look also at how this relates to the actual literature itself.

So since this is all in the same MCP, we can ask this information so we can search the literature for that specific topic of interest for clinical outcomes in obesity and type 2 diabetes, looking at papers using cites smart citation and citation graph of, 1.6 billion, over 1.6 billion smart citations that we've built to understand how the research is being discussed.

So this is going to go through and it's going to pull that information back.

We're going to get a sense for what the evidence base looks like as well as any papers that have significant contrasting citations that would potentially indicate that there's maybe an ongoing debate around that topic.

Now what's interesting about this is the MCP itself and what Claude is going to do as an LLM is it's going to go through and orchestrate its different queries using the Scite MCP based on what responses it's getting back.

So it went through and it made different comments and it made different calls based on what information was available.

So it built this evidence base for that specific intervention set.

It's showing all of the information here, it's showing the different steps that we're gone through and it's going to summarize all this information as we move through it.

But it was able to surface that and it's now telling about the citation count.

So understanding how the smart citations data is talking about the research specifically, and you can see that it's actually calling out the contrasting citations that are clustering and forming what would look like or signal a debate here.

So this is showing this thyroid cancer signal and it's talking about a specific set of papers here, or one paper that's related to that and it's talking about the tally.

So this is showing the number of contrasting citations, so supporting and contrasting information, showing that the contrasting citations are split.

So this is something to look into further.

Also it's surfacing another paper here related to this.

So talking about all this information and understanding how it works inside of the whole research landscape to see how things relate to one another, which is a power of what we're doing with Scite, and then offering that an MCP so that you can take this further.

So I could keep asking more questions or if you want to learn more about what you could potentially do with this.

On scitedemos.com, we have a list of artifacts that have been created out of all of the Scite MCP research that we've been doing to generate rich artifacts based on that data set.

So you can look through these and you can even look at specific prompts to generate these.

So if you were in Claude, you could look at papers and clinical trial information together and generate an interactive evidence map here, which is what I'm going to show over here.

So if you click on that, you're going to see a actual live HTML artifact that was generated with Claude.

This took five, five or 10 minutes, really.

This is comparatively, really small.

And you can even click around and you can look at the data.

we can look at all years.

We can look at preclinical, interventional, we can cite it down however much we want.

And it's showing all this information across 11 papers.

Five different trials, five different patent assignees.

So it's including even patent information, four different themes, and all of that rich context of smart citations via Scite.

So all of that is coming in here.

You can look at these individual papers and you can see those smart citation counts to get a sense for, you know, where debates are, how this has evolved over time.

And that's really what we're trying to do in the research landscape, is to help build on top of knowledge that's been created over time and see how that knowledge is evolving so that you can continue to fuel your research.

So there are many different ways you can use this.

Please let us know how the Scite MCP is working for you.

We're constantly improving it and trying to make it better.

So please reach out if you have any questions or feedback.

Thank you.