COLONTOWN Presents: Inside Look at ‘Second Look Cancer’ (2026)

Julie Clauer: [00:00:00] Hi everyone. Welcome to COLONTOWN Presents. Very, very, very excited to have, Dillan Eisenhaur here today with us.

He will tell you a little bit more about his story and about Second Look Cancer, of course, but I just want to say that to have a patient who goes through the trial search process and everything that’s painful about that and sees all the frustration and then takes that and actually does something about it, and not only does something but does something amazing about it, is really, really impressive.

So what Dillan has created out of his own experience and empathetically learn, listening to other patients, is created the first really CRC patient-centric AI tool, for searching for clinical trials. And it is amazing how significantly better it is than what existed for us as a tool, and I say as a tool because [00:01:00] obviously all of us are individuals and we know AI needs to be verified, so it doesn’t necessarily give you the answer perfectly without any intervention by you to figure out what’s good for you, but it really accelerates that process and really helps patients get to answers for them, much more quickly. And he has done all of this while having his own disease and also a day job.

So thank you Dillan, so much for your efforts in creating this and also for being here today to talk to us about it. So I’ll turn, turn it over to you.

Dillan Eisenhaur: Yeah. I’m very much looking, forward to talking about it and showing the tool off and you mentioned a disclaimer of sorts, but yeah, again, the entire website of Second Look is really built using AI and I think it’s a very powerful tool, but like any tool, AI or otherwise, it’s not a substitute for medical advice. So, at the end of the day, trial information should be verified with [00:02:00] some sort of care team. But nonetheless, I think the tool’s very powerful for its purposes, and I think it’s really industry leading for CRC. I don’t see any other tools that are at this point competing with it.

Julie Clauer: I’ll completely agree with that too, as an uninvolved perspective.

Dillan Eisenhaur: That’s good. Yeah, mine’s a little biased. Mine’s a little biased for sure. So yeah, I guess just to quickly, talk about the journey. So the idea to build Second Look came directly from using existing trial discovery tools as a stage IV patient. Throughout my journey, it was pretty much immediately clear to me that those tools were really limited from a patient’s perspective.

And that gap is really what led to me starting to build Second Look Cancer and just to go into a problem solution idea. The problem isn’t that trials don’t exist. There’s a lot of trials for CRC worldwide. [00:03:00] It’s that the infrastructure for finding and understanding those trials hasn’t really kept up with how patients make decisions in my opinion.

So Second Look was really built to — as the trial discovery infrastructure for patients specifically. So it doesn’t just provide a list of trials, it narrows down the trial landscape to options that are actually relevant for the patient based on their eligibility and location. Then it organizes and explains those options so patients can really understand the trials that they’re interested in.

So with that context, I’m happy to jump in and show off the platform a bit. It won’t be a full walkthrough of every feature. I think that would take a really long time. But the core idea of the platform can be shown off pretty quickly.

[00:04:00] Okay. So this is, the dashboard page is your hub but I’ll start with the user profile. So, Second Look is really built around the patient profile. It’s really the core critical aspect of the platform. And the idea is to progressively narrow down the trial landscape based on what actually drives eligibility, things like disease stage, biomarkers, location, and to show that filtering happening effectively in real time.

So, really quickly, if I go to my profile, I’m in Toronto, Canada, but let’s just say I was in New York. It automatically is going to show you the trials that are applicable in New York. So I’ll go back to Toronto just because that’s my actual location, and you can add multiple locations because the reality is patients are willing to travel or they have family in multiple places within the world.

But again, I won’t go through every single field here. [00:05:00] So, you put in your information, it’s relatively, most patients would know this information just offhand. And, the trial counter will just continuously update your filters, essentially. So what we’re doing is we’re scraping every single colorectal cancer trial, all of the eligibility criteria, and then we’re matching it to your profile and filtering things in that matter and filtering things out that don’t matter.

So again, I’ve put in, stage IV, where my mets were, and if I click “Next”, it continues to update, and, – same – you put in your treatments. And you put in your biomarkers, you click “Next” and it continues to update, and that gets you to a final number. So just to put it into perspective, Ontario has 80 clinical trials ongoing, and now we’re at 22, which are more applicable to me specifically.

So just from that, we’ve saved so much time of reviewing how many clinical trials that are not really important for me. And this is, – [00:06:00] one kind of example is ECOG which is like physical performance status. I’m a zero, but let’s just say I was a two. I have 22 applicable trials for me.

If I have an ECOG of two, there’s a lot of trials that only permit zero or one. So it’ll only update the ones that are actually relevant for you. So that’s just an example of, again, how powerful the system is. So that’s the user profile, that’s the core of the platform. Once we have that, we can send notifications on a daily basis of anything that happens that’s relevant to your profile. So if a new trial pops up tomorrow, I’m going to get a notification tomorrow that that trial is applicable to me. There’s a new Toronto clinical trial. So just having that profile, if you go on the website every day, you’re going to see updates that just automatically get sent to you.

So from here there’s really two [00:07:00] options and I’ll first go with Explore Trials, which, it’s more designed for patient control discovery. So this is for people that want to browse, filter, and understand trials on their own terms. While AI matching is really an AI tool that reviews all the trials that you’re applicable to, so these 22, for instance. And then it ranks them all, based on a bunch of different factors, which I’ll get into when we talk further about that. But just for the sake of looking at Explore Trials, this is the general idea where you get three cards, you get the trial card where the locations are.

Again, I’m in Canada, so it shows me Canadian locations. If you’re in the United States, it’ll show United States locations, European, different countries. It’ll show European locations, and then it’ll tell you how the treatment works. And there’s a very high level idea of it, and then there’s a more technical or scientific version.

So [00:08:00] really quickly you can already understand every single one of your trials, what it is and what it’s doing really quickly. And then the last card is the latest readouts and results. So if there is data online, we pull that and give you those metrics and source sources of them too. So, again, it’s another way of trying to determine your best trial relatively fast.

Instead of you having to go offline and search for readouts for every single trial, it’s right in the platform. ‘Cause for me, that’s one of the first things that I would want to know. And just to really put it into perspective, there’s a bunch of filters on the left side.

I’m a KRAS mutant, so if I only want to look at KRAS trials, I can just simply put in KRAS. And now these are the only trials that are applicable to me that are specific to the KRAS gene. And if I clear filters, just to again, hammer it in of how powerful the platform is, [00:09:00] there’s currently about 1600 trials for colorectal cancer.

And just by filtering for my profile now, there’s 22 that are actually applicable to me in my location, my stage, my biomarkers, et cetera. So that’s really the idea quickly of the Explore Trials page. There’s also a map function, which I will show you. It’s essentially Google Maps with a layer of clinical trials over it, of where the trials are, and you can see different hospitals versus just a trial list.

So really quickly, if I turn my profile off, it’s only going to show me when my profile is turned on. It’ll only show me ones that are applicable to my profile. But let’s just say I wanted to go to New York. It’ll show me all the ones in New York really quickly. So it’s just another easier way .

If you only want to look in a certain geography, there’s a very quick way of trying to find trials that way. So that’s the Explore [00:10:00] Trials.

Julie Clauer: I think, when you do it, it seems so, so obvious, of course it would do that, but I think those of us who have used a lot of different search tools, it’s like, “oh my God, it can do that!?” So if you’re new to search, just know that it’s awesome that you’re coming in with this as your, as your first, as your first option.

Dillan Eisenhaur: Yeah. Yeah. All right. So I’ll move to AI matching. So this might be a little bit more technical. I’ll try to trim it down as best I can, but the AI matching looks at the same eligibility logic, that, we currently have in your profile, these trials. But it’s really, instead of manual browsing, what happens when for me, I’ve already had, the AI matching.

I use the web. This is the only website I use for trial discovery but when you run it, all of your profile, excluding your name or your personal details, is sent [00:11:00] to our AI system. And, all the trials are, so your profile and all your trials, and then it tries to rank them, one through 10.

It gives you the top 10 trials that it thinks would be a good fit for you. And how it determines that fit, is through four different pillars. It’s biomarker strategy, treatment fit, scientific evidence, and logistical. And they’re all weighted differently. Biomarker is 50. Treatment fit’s 20, scientific evidence is 20, and logistical ease is 10.

So when you think of the algorithm, there’s four different buckets that determine the ranking. And I’m trying to make this as transparent as I possibly can because at the end of the day, this is really where, AI is new. Second Look is the only platform that does this at all, of ranking trials.

But I think it’s really important, particularly for people that, don’t want to look through 20 plus – some people have 80 trials, that are [00:12:00] applicable to them. This is a great way to trial discovery. One feature that will be included at some point is that you can actually guide the AI and what you want. So if I’m only interested in immunotherapy trials, I can tell it I’m only in interested in immunotherapy trials for the trials that are applicable to me, and it’ll send you the list really quickly. So, future things are coming with this, but, I think it’s still powerful as is.

So really quickly to go through the four pillars, biomarker strategy. Right now there’s really good drugs within the CRC trial space specific to biomarkers. So I’ve weighted that heavily. So when it gets a score, it’s going to be a pretty big part of the overall ranking. Same with any procedures.

They’re also ranked very high because you can theoretically get to “No Evidence of Disease” via different procedures or surgeries. Treatment [00:13:00] fit: this looks at your prior treatments and determines if the trial is doing something different or the same as one of your prior treatments.

So if you’ve already failed on an immunotherapy targeting something, it probably shouldn’t rank something that’s the exact same target. It’ll look for things that are different. That’s really the idea, scientific evidence at the core, this is going to rate trials that have more data on them than trials that don’t.

That’s really it. So a phase three is likely going to be rated higher in this category than a phase one with, with little to no data. and then logistical ease. This is a 10% weight. I think it’s very important still to have it as part of the formula because, let’s say some trials require, going to the hospital every day for two weeks.

Logistically that can be challenging for patients. So it’s a small weight, but it still matters and. [00:14:00] Again, going back to transparency, the AI says, our AI system says exactly why we rated each of these pillars. so trying to give as much transparency of what’s happening in the background to the foreground, to the patient is the idea.

And then all these are again, just pieces of the trial. So why the trial matches with your profile, what the drugs do. You can see a detailed description, which is direct from the trial listing, the eligibility requirements, et cetera, is all right in one place and you can even, share your matches and get a link to share your matches with people so it’s just a really cool way to find trials. That very innovative compared to what else is out there. So that’s an overview of the AI matching. I think it’ll only get better with time.

So again, [00:15:00] with all AI right now, I think it’s very strong. It’s a very strong tool right now, but this is the worst it’ll ever be. Even if you look at the website, we’ve only launched three months ago, the website is materially different from three months ago than what it is now.

It’s significantly better. And, we’re up to over 500 users within about three and some months. So, expect improvements all the time. But even as is, I think it’s just a really great tool.

Julie Clauer: Yeah, and I think on the match, I think what you said too about the transparency is so important because when you look at the criteria, right?

You decided on the algorithm based on what you know from patients and the broad overall general patient population, which is fantastic. But as we know with everything with general patient populations is, each of us is an individual.

And so with that, being able to see it, you might [00:16:00] say, “well, I don’t care as much about the biomarker part”. And then, because you can see where the things you care about so even though they’re weighted lower, you can see that, and then be able to find what you care about. And I also like that , as we know with AI, sometimes it gets things wrong, but this allows you to go to the source and actually investigate it, which is so, so very helpful when it comes to AI. And I agree, it’s improved dramatically and will continue to improve, but even on those glitches that exist, because of just the inherent nature of AI are also where people don’t have their own opinions.

It’s like this allows you to navigate that. So I love that as well.

Dillan Eisenhaur: Yeah, I definitely think it’s really important to be transparent and not have a black box of AI that people don’t understand what’s going on within it. I think, everybody gets a lot [00:17:00] more comfort when you can explain, or it can explain how it’s thinking.

And to me that was always important because again, as a patient, I don’t want just a list of one to 10. I want to know where that came from and why. Why number one was marked number one versus number five is number five. So that’s really the idea is to try to give people some context into what the system is doing.

Some smaller features, still relevant, but I would say those are the core features that we’ve discussed. User profile, explore trials, AI matching. Smaller features are: Save Trials. So these are all the trials that I’m personally tracking or interested in. And what Save Trials does is one, you can just quickly find it.

That’s the idea. But two, the second you save a trial, the system on a daily basis scrapes the trials and if any changes happen to this trial, so we’re talking new sites, [00:18:00] new information, new phases, if it goes from recruiting to, no longer recruiting, but active, anything that happens.

So that trial listing. You’ll be notified, and what I think is really important about that is if there’s a trial you’re interested in, but it’s not necessarily in the location, you would prefer the day that that changes and a new site is launched, let’s say right next to you, you’ll know. And I think that’s really powerful for patients to be able to be the first ones at trials to some degree.

You’ll be able to make those appointments as quick as possible if you know there’s a trial that’s near you, or at least be educated to be aware of that. So really quickly, this is really how it looks. Different completion dates, trial timelines, different hospital site changes, that’s really the idea behind it.

So I have a number of trials that I’m tracking and it’s been really, really useful. And, [00:19:00] likewise with your profile too, any new trials that are applicable are going to show up in this notifications bar. So if a trial gets launched tomorrow, you’re going to be notified of that trial eventually, there might may be email connectivity where you don’t have to log onto the platform, you’ll just get sent an email of, “Hey, there’s a new trial”.

Again, coming to the future, expect new and cool things where obviously that’ll be an opt-in email communication if you want it. But lots of feature ideas in the near future for sure.

Julie Clauer: Well, one thing I want to say, and I’m not trying to rain on the powerfulness of what you just described, but I also want to be clear about some of the limitations outside of this, right? Is that when you have saved trial and you have those notifications or just in general, even not on saved trials is all of this is based on the data that’s in public databases, like clinical trials.gov and that data is pretty good when it comes to a [00:20:00] new trial, not as good in terms of updates.

And so I’m just raising that so that people know that if there’s a trial that you are really interested in that isn’t just a single site trial, ’cause single site trials rarely go other places, but it’s in multiple places, but it’s not near you and you have it as a saved trial you’re watching.

That’s why communities like COLONTOWN are so important because having that discussion, because the lag in the source data is the problem. And so it’s a real reality.

Second Look Cancer will hit it as quickly as it hits that source data, but it might not be updated. There was a trial that I knew was open at a site and I kept talking to the research team about getting it updated on clinical trials.gov, and it was a 10 month process because everything needs to be approved and all, et cetera, et cetera.

So just as useful as it is, because it gives you those notifications, also use it as a way to remind yourself to ask [00:21:00] questions about any trials that are on there that are maybe of interest to you that you haven’t heard an update on in a while.

Dillan Eisenhaur: Yeah, it’s a really good point, like we’ve talked about the limitations of AI and I think, another limitation of any clinical trial matching website, mine or otherwise, is the data that we’re pulling from has to be accurate or that’s a limitation in itself.

If companies aren’t updating the trial data, then we’re limited in that aspect, because we can’t make good data. I’m pulling from primarily clinicaltrials.gov so that’s the reality of, how things are. And of course, I track a lot of trials and not all of them are perfect from a clinicaltrial.gov accuracy standpoint.

Julie Clauer: Right. Which is why this is such a powerful resource and tool. And you can use it as it’s intended, but you can also use it like in these other ways, which is, “Hey, I want to check in on my same trials. [00:22:00] And you, see what’s happening with other patients.

And if anybody knows anything about these trials, then you can use this as a source for that as well. So I think that it helps in terms of some of those other limitations that are outside of all of our control. But I think that’s why we’re never going to have a perfect tool because of these external limitations.

Dillan Eisenhaur: Yeah.

Julie Clauer: But if we can use the tool to get to our personal needs as effectively as possible, it’s really, really helpful.

Dillan Eisenhaur: Yeah. Yeah. So I think that’s a good overview of Second Look, happy to take questions.

Julie Clauer: So, question about the sources for the trial readouts.

Dillan Eisenhaur: Sources for trial readouts is primarily clinicaltrials.gov. That’s essentially my database, pulls in from clinicaltrials.gov and [00:23:00] is updated on a daily basis. So, any trial that changes – I’m aware. But yeah, that’s trials.gov.

Julie Clauer: What about though the results and the science behind the trial?

Dillan Eisenhaur: Oh, okay. So really quickly, so how the treatment works, I put a little AI symbol here, but that’s AI driven. So what the AI is doing is I send it the trial listing, and I effectively, I’m telling it to explain what the drug or therapy is doing in plain English and in a patient friendly sense. That’s really it. And then the latest readouts and results. This is an internet scrape. So similarly in AI goes to the internet. Not to get too technical, but there’s a ranking system based on source. So medical journals are going to be ranked a whole lot higher than, let’s say, a public internet forum.

Those aren’t [00:24:00] actually, – I won’t take those in. So, you take that ranking, and the top ones get, – PubMed is obviously a big one. So we take the response rates and such from those and put it into our system. So a lot of it’s AI driven.

Julie Clauer: Great. And can you talk a little bit more about, colorectal cancer trials?

One of the challenges with some of the searches is that solid tumor trials that include colorectal cancer, usually sometimes get ignored or they get pulled in, in not very helpful ways. Whereas it seems like Second Look cancer has addressed some of those challenges.

So can you talk a little bit about that, that when you say colorectal cancer trials, how that landscape looks?

Dillan Eisenhaur: Yeah. So that was one of the bigger challenges. The reality is, once again, like [00:25:00] those trials, they may say that they’re recruiting for colorectal cancer, but only certain sites are.

So it is a challenge. What I’ve done is, so anything tagged with colorectal cancer in general is going to get picked up by our system. Solid tumors has to go through an extra AI scrape to make sure that it’s actually specific to colorectal cancer. So if, if there has an arm that, just to give some background for people, if there’s an arm for colorectal cancer patients, so the trials actively including colorectal cancer patients, that would be included if there’s not, generally it’s excluded and that’s the way we’ve solved or attempted to solve the issue, it’s not going to be perfect. My perfect scenario would be that the sponsor that’s trying to get patients in the colorectal cancer space would tag that trial as colorectal cancer [00:26:00] rather than just a solid tumor trial.

In my opinion, solid tumors shouldn’t be an option in clinicaltrials.gov tagging. You should just put the cancers you’re actually looking for within the trial. My take.

Julie Clauer: So if it’s a solid tumor trial, but they’ve tagged colorectal cancer and it doesn’t necessarily have a colorectal cancer specific arm that would be included.

Dillan Eisenhaur: Yeah. If they’ve tagged it, it’ll be included, which I would imagine if they’re tagging the trial as colorectal cancer, there’s probably an arm or there’s an interest in getting patients within that trial for that. If they haven’t tagged it, that’s where there’s an additional AI scrape to try to make sure, to some degree of accuracy that it’s applicable.

Because I think one thing I want to try to avoid with the database is getting trials within our website that aren’t for colorectal cancer. And I think this is another point of I don’t want to say ‘edge’ for Second Look, but [00:27:00] since it’s dedicated to colorectal cancer, you can make those decisions versus other websites that may give you trials, or more trials that are not actually applicable to you because it’s generally all pan-cancer based.

There’s no other trial finders that I’m aware of or that are generally larger, that are specific to colorectal cancer. They’re all different cancer types. So that’ll generally get you more matches and some of those matches won’t be actually applicable for you.

Julie Clauer: Great. Indeed.

So let’s talk a little bit about the international applicability. Obviously as a Canadian, we know that this is bigger than US-based. It actually started with Canada and then you’ve expanded it. But I think especially for places like Europe where patients are looking across multiple countries, I know that’s an option.

But also in terms of the source, because there are quite [00:28:00] a few trials internationally that aren’t necessarily listed on clinicaltrials.gov, and are on other databases. So, I know that that’s a big ask, but is that something that’s in the plans right now?

Dillan Eisenhaur: So I’m aware of those databases.

Getting them hooked up is going to take some time. There’s some technical things to think about while doing it. It will happen before June to have those extra databases within there. My assumption it’s about 10 to 20% at the high end of trials that could be missing from Europe, so it’s not a significant amount, but obviously we want a full gamut.

We want every single trial possible, so it’s going to happen. It’s a priority. The real “if” within my life, it’s always what’s the next priority and that’s on there. And then a mobile app is also on there, because right now about 60% of all users are logging in on their phone. So the idea of an app I think [00:29:00] resonates with a lot of people.

But, doing that means I’m not doing other things first. So it’s managing priorities and top level things.

Julie Clauer: Just FYI I only use it on my mobile, and the mobile friendly web is fan-fricking-tastic.

Dillan Eisenhaur: That was fully redesigned last month and I’m really, really happy with it now. So. Yeah.

Julie Clauer: So, so you got a lot of bang for your buck in that. So question, this commenter is saying a huge thanks to you for creating such a tool. Can you tell us what guard rails exist to prevent the AI from generating plausible sounding, but incorrect matches?

Dillan Eisenhaur: Yeah. So the benefit with that is I guess just to give some background and try not to be super technical, but every single clinical trial listing has eligibility criteria.

I scrape every single one of those listings and create data points. So think of what [00:30:00] stage, how many lines of treatment, ECOG, these are all different data points. So when you think of the algorithm, it’s very easy for AI. All I’m asking the AI to do, and the first layer is to tell me how many lines of treatment this trial requires.

AI are really good at doing that right now. I don’t see an issue or an ability for it to give wrong trials. so that’s the source of truth when you think of it. From there it’s putting in your patient profile, which is completely accurate because it’s yours, you’ve designed it. And then we’re just matching those two things together.

So from a trial perspective, it’s going to be very, very accurate. Now with that said, I only pick up a subset of every single eligibility criteria and it’s getting better. But of course, some trials are going to have very atypical things that isn’t realistic to capture within a small patient profile. So the goal is to filter out 80 to [00:31:00] 90% of trials, not get it down to every single perfect one from a clinical accuracy perspective. But, having that source of truth, being clinicaltrials.gov and using an AI layer to parse it, it’s a very, very low risk of having inaccurate trials being sent to you. AI is really good when there’s databases and sources of the truth.

Not to get too technical again, but when there’s a source of truth and a database behind it, it’s very easy because you can limit it to certain scopes. I think a lot of times when inaccuracies happen, it’s when you give it too broad of a landscape. So if you allow internet access, that gives it a whole world of different data points that it can pull from.

And that’s doing “a-b” testing. That’s really where you see a lot of the inaccuracies come from, not from [00:32:00] specific database changes. So that’s my 2 cents.

Julie Clauer: Well, and it sounds like what you just described makes total sense to me for the search part of it, but when you get to the matching part of it where you’re ranking things based on scientific evidence for instance.

Dillan Eisenhaur: Okay. Yeah.

Julie Clauer: The accuracy of that scientific evidence, making it sound plausible, like it _could_ make sense, but then when you actually, if you put a…, somebody who’s a scientist that looks at it and says, “this actually isn’t at all really evidence for what you’re talking about”, I think that piece is a question.

Dillan Eisenhaur: Yeah, to be super honest it’s a limitation. AI’s won’t be perfect. There’s a number of things that I do to try to make it as accurate as possible. So for instance, pick up colorectal cancer, from scientific evidence journals, right? Don’t use public internet forums.

There’s ways to rank what source material to get the the best accurate [00:33:00] information. But the reality is, it can make mistakes, so that’s why, particularly with the AI matches, I would recommend having somebody in the medical field, your oncologist review it because there are realities that could happen. There are guardrails. It’d be really technical and it’d take a long time to put in how many guardrails I’ve put in, but to put in perspective, there’s about 800 lines of code that I’ve put in around guardrails for the AI to give it direction of what’s okay and what’s not okay but regardless, no tool is going to be perfect, and that’s a reality of trying to give it the best direction. But it’s even with a person, right? If you give 10 people directions, they’re going to come out with different listings. So that’s just a reality based on the AI limitation.

Julie Clauer: Well, [00:34:00] I will say too that you look at a biomarker test result, right? And those typically have trials recommendations in them. They’ll say, “here’s a trial potentially for you” and that’s from huge companies and whatnot. But I think because they’re so broad.

They are, I would say, just from my personal experience, significantly less accurate than what you’re doing. And so I do think it is a limitation for sure, and I think being transparent about those limitations and having people verify the whole verification is so critical.

But just to put it in context of AI matching tools from much more billion dollar companies, like bigger billion dollar companies and not Dillan doing this as his side gig, probably are more limited even on that. So just a note.

Dillan Eisenhaur: It’s a known limitation. I try to make it as low of a [00:35:00] risk as possible, but the risk still exists.

Julie Clauer: Okay, so we have a question about what are the MRD trials, and obviously maybe that person specifically wants to understand answers, which you can go to secondlookcancer.com and find them. But I would love, if you don’t mind talking about how if somebody does have, MRD stands for Minimal Residual Disease, which means that they don’t have anything on scan, but it shows up in their CT DNA test results.

So it’s in their bloodstream. We know it’s somewhere, but it’s not showing as a tumor yet. That’s a very specific group of patients who are very interested in trials. I think, you might be familiar with this and so what, and how would somebody like that approach, their profile and the search, to find those?

Dillan Eisenhaur: Yeah. I have a separate, – so I would fill out your profile like normal. If you don’t have mets from a scan, leave that section empty, but [00:36:00] really MRD trials are so different than any other trial, and they’re new, there’s less than 15 or 20 across the world right now. So, it’s a separate page because of that, it’s just too different that everything has to be different about them.

So right now, if you go on your dashboard page, there’s a big card that says ‘Explore MRD Trials’, and if you put in your profile effectively, it’ll filter for location. But otherwise, it’ll do stage and location, but I don’t want it to be too restrictive because realistically, there’s like 12 trials, right?

Where, the whole point of Second Look is to narrow down trials. But if there’s only 12, I think you should probably just look at them all to some extent. So that’s really it. It’s a separate page. I’m really excited about this area of clinical trials. It’s definitely where I’m most interested right now for [00:37:00] my personal side of things.

So I would recommend going on there. We keep them very accurate. I review these personally and there’s some patient advocates that I also work with that, we try to keep these as accurate as possible because it’s an ever-changing landscape. But really there’s only two hospitals that I’m aware of that are really moving these forward, and I expect that to grow over the next couple of years.

Julie Clauer: That’s great. So the answer for, if you are looking for MRD trials, there is a special “escape hatch” button or tab that you could push to find those. So thank you very much.

Dillan Eisenhaur: Secondlookcancer.com/mrd-trials. We’ll pull it up. You don’t even need an account to look at them because realistically it’s 12 trials.

Julie Clauer: That’s great. Thank you so much for that. So then another question out of pure curiosity, what LLM frameworks are you using and what is the anonymization protocol used to safeguard patient info? [00:38:00]

Dillan Eisenhaur: Okay. Very technical. Yeah, so generally speaking, nobody should be building their own models at this point.

They should be using models from one of the big AI companies because it costs hundreds of millions of dollars to really build something at an equal scale and impact. So I’m relatively indifferent from a model perspective on using open AI versus Google, versus Claude. I think all three are really, really good.

I go back and forth between them. Every few months I’ll do testing based on when they release their newest models. I don’t want to say which one we currently use because it can ever change. And we use AI in multiple different areas and I’ve used different models for different areas, like for search, Open AI and Google are really good.

Claude not so much in my opinion, but for parsing, I’ve [00:39:00] used all three. So, in general, all three work, as long as you’re using the most new and best models, generally they’re the most expensive too. Those are really the best to go with, from a security standpoint. All of them have what’s called, again, not to be too technical, but, private networks where the AI is housed within your cloud environment. So think of your server. Your server is not going outside of it and connecting to a public large language model, it’s contained within your environment. So there’s no ability for Google or Open AI to train the models based on that. They can’t research that data. It’s a literal model housed and running in your own environment, in your server. That’s the way all large companies, corporate or otherwise are currently doing it. And to me it’s the safest because there’s no [00:40:00] real ability for data leakage at that point.

If it’s not leaving your server or a little, theoretically Google or OpenAI, whatever server you’re using. They have some level of access to it, but it’s generally encrypted even at their level. So, yeah, hopefully, again, I don’t want to be too, too technical, but that’s the idea. Keep it inside your server.

Julie Clauer: Great, thank you. Should we be updating any AI results after big conferences like ASCO or ASCO GI? I think that maybe means the match, like go and check match.

Dillan Eisenhaur: Yeah, so the latest readouts aren’t updated on a daily basis. There is a date of when they’re updated, so right now I wouldn’t say it’s critical to do that.

In the future, it’s not a terrible idea to have all the readouts updated post an ASCO GI. So then [00:41:00] we have that data included, and in which case, refreshing your AI matches would make a whole lot of sense. So whoever sent that question, I think I have a new development item on my list, but right now I wouldn’t say it’s super important.

You could get, if a new trial was updated post ASCO GI that would be included in the AI matching. But, systematically, it’s not captured right now, right after an ASCO GI.

Julie Clauer: How often do you do up those updates?

Dillan Eisenhaur: Generally quarterly. So realistically they haven’t really been updated, that often since the launch of Second Look.

Usually it’s every quarter. I look at all of them together and look at the system instructions that are given and see if they need to be updated to be more accurate. And then, we do a full scrape on [00:42:00] every single colorectal cancer trial. So it’s a process that I would only ever want to do every three months, at the shortest span.

Julie Clauer: Yeah. That’s great. In the future, do you think it would be possible for a patient to exclude trials based on, for example, certain drugs, if a patient had a allergic reaction to “oxali”, for instance, or not, those kinds of things?

Dillan Eisenhaur: Yeah, yeah, absolutely possible. It’s not necessarily in my pipeline in the next three to six months, but I think this will be. Eventually there’s going to be a preference tab associated with your profile, which will have a bunch of questions that are really not about your specifics, but your preferences.

So, one great example that you gave me, Julie, is control arms where if there is a chance that you could get, the standard of care rather than the trial drug, I think that’s a preference that some people really care about, me [00:43:00] included. So having a preference table that says, “don’t even show me trials that have this”.

I think that’s where I would include if there’s allergic reactions to certain drugs, just take them all out. And it goes back to the overall concept and the point of Second Look is let’s filter out trials that you don’t want to see. So that aligns perfectly with the website.

Julie Clauer: Great. So I feel like, that the tool is very intuitive, very easy to use, but I also, have eight years of experience of living with this disease and looking for trials. Yeah. So, so let’s say somebody goes in there and they are just like so overwhelmed, they’ve just been diagnosed. They don’t understand any of the lingo.

What would you tell them? Like they fill out their profile? Then what would you say is the easiest way to know what to do next? Like they have the search results. Are they matching, do you think, [00:44:00] sending them to their doctor, do you think looking at the first one and seeing if it makes sense, what do you think?

Dillan Eisenhaur: Yeah, it’s a challenge because the reality is patients are on a spectrum of how much of a quarterback they want to be in their own care to some extent. So there’s a few things I would say to do. First off is, I think speaking with your doctor on the trials that you were given, is a great option. Speaking with others like a COLONTOWN or other support networks.

I think that it’s really good to get an idea of what other people are thinking, and then also just talking to the trials and contacting sites, having appointments and seeing what you’re interested in. But realistically, there is this is a great discovery tool. It’s not fantastic on going from discovery to actually getting into the trial yet.

That is a bigger lift that I’m going to be making easier on people. [00:45:00] But realistically the best option if you want to get into a trial is book an appointment at that hospital for the trial. But if you’re not ready to do that yet, then I think, speaking with your oncologist about it, speaking with others particularly that are interested in clinical trials are really good options.

I feel like you could almost answer that question better than me, to be honest.

Julie Clauer: No, not necessarily. I mean, there’s lots of, and we have searching which we’re actually reworking now because of Second Lung Cancer. So, so that that’s an option too. But I would say, posting in COLONTOWN would be what I would say. And say, “anybody know anything? Can we talk about this?” That’s the best way to learn, to me, is to ask other patients and see what they think. They know, because those of us that have been there can help share what we know and those who haven’t can learn together, you know? So, yeah, that’s my opinion, but I’m biased in my opinion on that, obviously.

And it doesn’t have to be COLONTOWN, any patient network, but I [00:46:00] think COLONTOWN’s very set up for that discussion. Okay. So somebody that missed the beginning, Dillan did talk briefly about his origin. This came from his own experience, but I would love for you to elaborate if you can, have a few minutes just to elaborate on your background and how now you were able to take that problem and actually get to the solution.

And then also, if you’re doing this on your own or how you’re coaching them.

Dillan Eisenhaur: Yeah. So, background, I’m not technically in technology. I’m in finance. I work. My day job is at a large Canadian pension plan, on the investment side of things.

I’ve always been interested in coding and technology in general, just because even in that role, you need technology to really, like a lot of my role was analytical. You need really good analytical tools to [00:47:00] perform well realistically. Second Look is really just a lot of analytics put together in different ways. So, it really became an easy concept and I’ve learned coding throughout my career. So, it was a relatively easy lift, because at the end of the day, all we’re doing is manipulating trials from one database. It’s actually much easier than a lot of tools out there, right?

From an investment standpoint, stocks, there’s so many different sources, of information to gather from. So it’s one information source, which is great, much easier on me. From a coding perspective, it’s only me, so far. I think that could change in the future depending on how Second Look grows and the demand for it and how big the user base gets.

I think there’s a scale where more people will be involved, from features and beta testing. There’s lots of others [00:48:00] involved, which I really, really appreciate. I, COLONTOWN has been one of the the folks or Julie in particular, of giving feedback of different ways to implement features or show different information and patient advocate and Delores has also been really, really helpful in beta testing, and giving feature ideas. But in general, Second Look is really a community tool and the best way for it to grow and be helpful for patients is getting patient feedback: what’s working, what’s not working, what do you wish the app could do that it’s currently not, that’s all going to be patient-driven at the end of the day.

Because the idea of it is to be a patient focused tool. The fortunate part is I’m also patient, so I have an idea of what I want to see out of a tool, so that’s really been a huge driver of the entire website.

Julie Clauer: And obviously you are very open and responsive to people that reach out to you, but what is the best way, [00:49:00] is there like a feedback mechanism tool on Second Look Cancer?

Or what’s the best way for somebody to do that, to reach out to you if they’re not connected to you in other ways?

Dillan Eisenhaur: Yeah, the best way is, [email protected]. if you are on the website, if you scroll to the bottom, it’s in the footer. Feel free to message me. Again, I don’t think there’s been an email yet that I haven’t responded to within 24 hours.

I’m online every day working on this, just out of a passion for it. So, if people ever have feedback bugs, thoughts, thanks. I’m always happy to hear it.

Julie Clauer: Awesome. So somebody raised a question, which I’m going to edit it a little bit because part of it, you answered earlier, before they were able to join, so they can go back and watch that part.

But it is a question: if somebody’s like, “I don’t want to create a profile. I don’t want to become a member of something”. What is the value for them, basically, if they don’t want to do that.

Dillan Eisenhaur: It’s tough. In [00:50:00] general, the whole point of Second Look is to build a patient profile, so then it’s geared towards your specific situation. I’m aware of that. I’m the same way. I don’t like making accounts on things. You pretty much have to make an account. Now, if you don’t want to put a patient profile in, you can use the Explore Trials tool to use the filters from, geography to stage to biomarkers but I will say the platform’s built on the patient profile, and the real value proposition for patients is to fill it in to get all of the powerful tools that are available with it. So it’s an unfortunate reality that I built it around that. So it’s a, I guess a limitation, but also by design.

Julie Clauer: Annie raised a good point that it’s free. So yeah, I understand. All of this is free. We did not mention that. So thank you for raising that. Dillan is very passionate about this, as you can see in terms of helping patients. And [00:51:00] it is all free to patients to use, and provide their 2 cents to Dillan.

He takes feedback for free too. It is interesting though and helpful to know that you can use the search filters if you do want to just see trials, but if you’re concerned about creating a profile, you do have the power of the search filters, even though it’s –

Dillan Eisenhaur: About 7% of users are doing that so it’s a small minority, which makes sense. But people do, do that and that’s totally fine. It’s whatever works for you individually.

Julie Clauer: Yeah. Great. How is the project being funded?

Dillan Eisenhaur: I’m funding it. There’s really no immediate plans to change that. The real question will be in the future if this platform becomes significant, and I mean like a meaningful amount of colorectal cancer patients are part of the website.

And, at that point that the question is how big can it get and what cool features do we want to include, which I will either have [00:52:00] to fund myself or find funding for. The reality is it’ll always be free for patients. there’s never, and I have no plans to ever change that or will ever change that.

So right now, for the foreseeable future, it’s just solo funded and that’s what it is.

Julie Clauer: Well, and I think too, that funding sources, I think part of it is, understanding. What biases might be in the tool from funding sources. So knowing that it’s not, but I know also that you are very cognizant of that in that some search tools are funded by sponsors of trials, for instance.

And in that regard, those trials might come up first in a search, and so that is something we tell people to look for is, what is the funding source so that you can be aware of that. So just an education tip on just something to look for in search tools in general.

Dillan Eisenhaur: Yeah. [00:53:00] My view just on that point, my view in general is transparency is very important here, where if we were to get into a partnership with anybody not-for-profit or a pharma company, we’d have a partnership tab on our website discussing the partnerships. From a theoretical standpoint, I would never change the trial matching algorithm based on some source, and I wouldn’t even have funding if that meant an appearance of that conflict of interest. So from that standpoint, it’s very limiting. I just couldn’t do it. So not-for-profit partnering would make a whole lot more sense for me than pharma per se.

So yeah, I definitely think it’s a longer conversation that I can have over that. But in general, the quick answer is I’m not really planning to do anything else, and be funded myself right now.

Julie Clauer: Thank you. What about trials that don’t involve experimental drugs? Are those included as well? So – things I think that aren’t necessarily, well, first of all, [00:54:00] treatments that aren’t necessarily drug like, like, local interventions, right? But then also there’s a lot of trials that are supportive, care related or preventative or those kinds of things.

Would those be in there?

Dillan Eisenhaur: Yeah. So how I designed the website so far is that, there’s interventional trials, which are, you’re doing something to directly affect your situation. And then there’s observational trials, which are more studies or monitoring or seeing how things play out in the future.

But there’s no actual treatment per se that you’re undergoing. Those aren’t included in the website. That could be a future version if people are interested in those, but I think the mass majority of patients that are trying to find clinical trials are trying to find clinical trials where some form of treatment is happening.

So right now it’s all of those, it does include procedures, it includes different drugs or, otherwise, and there’s [00:55:00] filters that you can use to specifically look at procedures versus drugs, for instance. But it only is for things that actually are changing. You’re undergoing some treatment or procedure.

Julie Clauer: But those interventional trials on clinicaltrials.gov, those do include things that aren’t necessarily for treating the cancer.

So, for instance, it might be a side effect treatment or it could be the aspirin to prevent recurrence. Right. Those kinds of things. So those would be included.

Dillan Eisenhaur: Because they’re part of those interventional trials. Yeah. If it’s listed as interventional, it’ll be included.

Julie Clauer: Great. (reading a comment that was posted) …repurpose drugs… or two FDA approved drugs in a new combination or move to earlier line of treatment… yeah, that was an answer to that question. So yeah, I think all of those would be part of interventional.

Dillan Eisenhaur: Yeah. If it’s a trial now, if it’s, what’s the terminology? [00:56:00] I forget the terminology, sorry.

Where, where you’re not actually undergoing a trial, but you’re getting it for off-label use or something of that sort. That’s not part the website.

Julie Clauer: Yeah. Also t hings like histotripsy or brachytherapy for rectal cancer or watch and wait for rectal cancer. So all of those would be in the database as interventional trials. So they would be, part of this.

Oncologists can sign up. You don’t have to necessarily be a patient to sign up. It’s not like they’re looking at your patient credentials, ’cause, caregivers can sign up, patients can sign up, oncologists can sign up. I think it’s actually a fantastic tool for oncologists because they have just as much trouble as we do sometimes understanding what are all the trials available for their patients.

But the design is designed for, patients.

Dillan Eisenhaur: Yeah, that’s precisely it. The patient, I’m happy with anybody signing up if anybody wants to make an account oncologist, hospital or otherwise, happy to have you.

Julie Clauer: Right? Totally. Yes. ’cause it’s like our DocTalks. We find that actually [00:57:00] oncologists really enjoy our  DocTalks because they learn from them even though they’re intended for our patient audience.

Dillan Eisenhaur: Yeah.

Julie Clauer: Any other final questions? Thank you so much, Dillan, for, again, for creating this and then also for your willingness to continue to work and be so passionate to make it better and better and really making an impact for those of us who need to be in the clinical trials space.

Dillan Eisenhaur: Yeah. Thanks.

Thanks so much for having me. I’m definitely, really excited about trying to move the needle a little bit in terms of clinical trials and I’m laser focused on them. Very excited to have -I have many, many future ideas on what the app can look like and expand to. So, I hope everybody stays tuned and sees where this can go.

Julie Clauer: Yes, and I actually have an idea that, of something that I want to use it for within COLONTOWN. So keep your eyes open, COLONTOWN members for [00:58:00] something in the works. Okay.

Dillan Eisenhaur: Sounds good.

Julie Clauer: Thank you so much, Dillan. Thank you everybody for joining. If you’re watching this on recording, if you have questions, you can send them to supportatsecondlookcancer.com and thank you very much.

Bye.

Dillan Eisenhaur: Of course. Thanks so much.