Aiforia Technologies Oyj (HEL:AIFORIA)
Finland flag Finland · Delayed Price · Currency is EUR
0.8700
-0.0300 (-3.33%)
Nov 17, 2026, 6:25 PM EET
← View all transcripts

Investor update

Nov 17, 2026

Summary

The digital pathology market is forecast to grow 12% YoY through 2031, while clinical AI adoption remains early. 11 regulated solutions and 20+ clinical customers support expansion, but implementation delays and reimbursement uncertainty slow revenue conversion.

Jukka Tapaninen
CEO, Aiforia Technologies

Good morning, and welcome to Aiforia's investor webinar. My name is Jukka Tapaninen. I am the CEO of Aiforia, and today we have two keynote speakers, and at the end of those, I will have my short presentation about Aiforia and Aiforia's situation. Two speakers and two perspectives. One is the first one, Professor Angelo Paolo Dei Tos from Padua, Italy, and Paolo is a key opinion leader in this industry. He is a full professor of pathology and dean of the University of Padua School of Medicine. He has been, of course, part of the many kind of associations, but he is a faculty member since 2000 in the European Society for Medical Oncology, ESMO. He has been writing more than 580 scientific papers on these topics, and he is the president of Italian Society of Pathology.

From his side, we will get the view of the science side, plus the view of the user and their view on the pathology and AI and diagnostics overall. The second speaker, we have Katie Maloney. Katie is a partner at DeciBio, a precision medicine strategy firm. She is based in London, actually leads the London office, and she is a co-author of the Digital and Computational Pathology Market Report. From her side, we are hearing what is going on in the market, what is going on with the digitization, what is going on with AI-assisted diagnostics, and so on. At this point of the time, I would really like to kind of hand over to Professor Dei Tos to start the first kind of presentation, and after that, we have a short time for questions. Please, Professor Dei Tos, please continue.

Angelo Paolo Dei Tos
Professor of Pathology and Dean of School of Medicine, University of Padua

Thank you, Jukka, for such a kind introduction. Also, thank you for inviting me, and I wish to greet all the people who are part of this meeting. Just briefly, the point is whether pathology and AI is how it is doing, going together, if we are ready for that, and I try just to share with you some of the views, current views on this. As you know, the story of AI is a very long story, by the way. Now, of course, everybody is excited about the use of AI. Some people is also, I mean, is kind of worried about the use of AI, but I mean, the basic question is whether we can get help from this new technology and just to improve the quality of what we do every day. Of course, the impact of AI on medicine is huge.

Today, we are talking about diagnosis and, to some extent, also imaging. Of course, I believe that this is very, very important to make precision medicine or personalized medicine becoming a reality. Of course, the possibility to use AI spans from patient management, drug discovery, robotics, telemedicine, and remote assistance. The management of chronic disease, there are many countries in the world now that, for example, they are using AI to monitor distance hypertension or diabetes. Of course, the basic idea is how we can use this technology in order to improve the quality of our decision processing.

If you go to any of the big cancer or medicine journals, you will see how AI now is very pervasive, and basically, there is a huge amount of very, very top-notch papers coming out, trying to, in a way, to underline how we can make a nice use of this new technology. What we do basically nowadays is changing. At the very beginning was morphology, then, of course, we use some techniques like immunohistochemistry just to realize what type of the tumor is, and then we came to a diagnosis, then to a treatment. Now, we had to add something more, which is basically molecular subtyping, which brings into actionable biology, which is exploiting some of the genetic or epigenomic alteration in cancer mostly to use specific drugs. This is going to be integrating therapy selections.

Actually, in this scenario, pathology doesn't make diagnosis, so you need to know what disease is present. Then it would be prognostic, so what outcome is likely to expect in this specific patient, but also predictive, which is a nicer part of the story, which treatment may help. This is what we do. That's why I believe pathology is the cornerstone of clinical decision-making. So, pathology nowadays, for the people who doesn't know what is pathology, is the core of any clinical decision-making, particularly in oncology. What is next? Next is the wish to integrate all the data we have. Now, of course, we're starting discussing about big data.

We are not satisfied where we are, because when we talk about precision medicine, okay, we can adjust some specific drugs to very specific genetic alterations, but we are not at the point in which, for example, we can integrate all what we know about specific cancer, all what we know about the specific human being, all what we know about the specific genomic or epigenomic expressions, all what we know about the imaging. To do that, we need AI. Pathology, in particular, to be part of this process, need to go two main steps. The first steps is make an analogic, a live imaging into a digital slides, and the second point is to use AI. To do what? To do patient risk stratifications, individualized patient risk stratification. Without AI, we will never achieve this type of results.

Basically, pathology is now living a kind of a magic moment in which we need to be proud of what we did in the past, but we need to look at the future. The future now is a process of digitalization of the imaging, but the only reason to leave the microscope and to make digital slides, which is an important investment that we are doing currently in pathology globally, the only reason to go digital is to enable AI. Otherwise, we could easily stay on what we have now, which is beautiful microscopes. In Veneto, which is my region, is a region that in a way live like five million people, we are trying to implement this process. Why we're doing that? Because there are many situation in which AI can help in daily practice.

For example, improve workflow and reduce turnaround time, which seem to be very relevant for clinicians at the moment. From the pathology standpoint, to level the playing field, and so to increase accuracy, to reduce inter-observer variability. Also, quality assurance issues like secondary reads before final sign-out in order to ensure that clinical significant lesion have not been missed. Most importantly, do things we don't like doing anymore, like counting cells and so on, all these kind of things that AI can do far better than we do, but also do things we can't do, because some subtleties can only be picked up by this new algorithm. Within an image, a digital image, there is so many information, we already know this, that can be so relevant in terms to achieve the main aim, which is personalized medicine.

As I mentioned to you, we live in Veneto, five million people. We have two academic hospitals plus five network hospitals. We generate approximately three million slides per year. What we did basically at the very beginning is to, and this was done mostly with the funding of PNRR, and what we did was to achieve a common platform with a common LIS for all the departments of pathology in my regions. The good thing that we could do at the time with that money is also to implement AI in our diagnostic process. This is basically my region. This is the distribution of the many hospitals that we have. Basically, we did one LIS, artificial intelligence, and Aiforia, and this was through a tender, was the partner that we selected.

What we are doing with Aiforia currently, but we are about to expand further next year our collaborations, is a support for prostate biopsy screening and Gleason score, and we show you some data. Automated evaluation for breast cancer immunobiomarker, like ER, PR, and Ki-67, and now HER2 is coming. Also very importantly, computer-assisted PD-L1 scoring. PD-L1 is the biomarker that we use for immunotherapy in many type of cancer, like lung cancer, like melanoma, like GYN cancer, and so on. This is very important to us. The reason we are using AI now, because we already know that we improves by 20% the quality of the prediction power of PD-L1, which means basically admitting to immunotherapy, which is very effective, 20% more patient than we did in the past. Why prostate?

Well, if you look, go to the literature, the prostate is kind of skyrocketing in terms of use of AI, and the reason is that when we receive prostatic biopsy, this is what happens when you're using the Aiforia technology, you have to evaluate, let's say, 12, 24 core biopsy, is very time-consuming. The system helps basically in recognizing the grading according to Gleason. What is a grading? Is a parameter that tell you what is the likelihood of the tumor to be aggressive. This parameter impact on the treatment. For example, to decide whether I'm going to do surgery or not doing surgery at all.

And what is interesting of the use in prostate pathology is actually that not only we could see quite accurately what the Gleason score is, and so having a great support in terms of clinical decisions, but also an excellent support for diagnosis. Basically, the negative predictive value of the system is 99%, better of any pathology in the world, which means that if the system basically doesn't recognize a cancer, which is the prerequisite to do the Gleason score, then you're pretty sure that this case is negative. This is one of the amazing application of AI to pathology, because as a screening tool, because screening these cases and many other cases, many other situation, is extremely time-consuming. This will be fantastic to persuade people to use and to trust the system, because we already did all the validations, and it works very, very nicely.

And what about the future? Well, as I mentioned to you, one of the aim to get to fully achieve precision medicine. Until now, we have been focusing on the specific cancer, but we are not focusing on where the specific cancer is sitting, what are the interaction of the cancer with normal cells, with immune system, with the stroma, and so on. The only way to, in a way, go to a tumor ecosystems instead of simply focusing on a specific type of cancer represent the next frontier. I'm talking about the spatial techniques, spatial transcriptomics that we can use in cancer research, in neurosciences, so it's just not pathology. Developmental biology, disease pathology, of course, even non-neoplastic. I will just give a couple of examples.

And the baseline, you can't do anything like this without AI, because the number of data you generate can be only sorted out with the use of AI algorithms. This is one example. The idea is, how can we tell from a digital slide whether the clinical outcome will be good or bad? Is it possible? Just one example. These are two breast cancer. According to the normal biomarkers that we evaluate in pathology, they are identical. But you know from the colors, one is very small, azure and blue, and the other is very red, which means good and bad, try to be very simple. So, actually, spatial transcriptomics based on AI can give you predictive and prognostic information, which are almost impossible to get when we do routine pathology. I'm not telling you that this is the routine now. It's very expensive still.

We are doing this on research, but it's very advanced research. So I believe that in the range of three to four years, you'll be using this type of data and this type of technology even in a clinical context. Of course, we know that there are two phases of AI, so there is many discussion currently about the potential dangers. We are all facing this situation. But we need to consider that there is a potential there. So in pathology, we are simply using AI as enhancement of human potential. We don't want to become apathetic or lose motivations. This is part of progress and innovations. And to use properly, really to have a discovery mindset, which is part of medicine and pathology in particular.

I believe that this technology, if used properly, represents really, really an amazing tool in order to shape the future of medicine and the future of pathology. AI, it is undoubted that represents a revolutionary tool. Any field of medicine can be involved. Pathology, I believe, is currently representing one of the best fields of application. The process, of course, is slow at the moment, but you have to consider that pathologists are very conservative, so you need to be persuaded. You need to understand that implementation of digital technology takes time, but also reaching the break-even on the investment takes like five or six years. That is why the process seems to be slow, but it is going on, and everybody now is understanding the value which is in this new approach. Of course, precision oncology needs big data analysis. You cannot do anything without AI.

Of course, we need transparency and ethical rules. Nobody likes the black box, so it is important that when we do apply algorithm in a way we understand the principle behind the choices that have been made to build the algorithm. We need to know what is going to evaluate, because this increases the trust of the people when using these tools. Of course, one of the issues is global commitment to inclusive applications, because we all know that we live in a world of rich inequalities. AI can be an opportunity, but we need also to avoid it becoming a way, a mean to further broaden the gaps between the high-income countries and low-income countries. Having said that, I wish to thank you for your attention, and of course, later on, I will be happy to answer any questions.

Moderator

Thank you, Professor Dei Tos, for your presentation. Before we jump to Katie's part, there are a few questions from the audience for you.

Angelo Paolo Dei Tos
Professor of Pathology and Dean of School of Medicine, University of Padua

Yeah.

Moderator

Can you put numbers on the benefits so far? For example, turnaround times, pathologist workload, or consistency between readers?

Angelo Paolo Dei Tos
Professor of Pathology and Dean of School of Medicine, University of Padua

Yeah. Now, actually, what we are doing in Veneto, and this is a public system. Basically now, we are using this technology in four places, Padua and Verona, which are the two academic hospitals, and another two spoke hospital, which is Vicenza and Rovigo. The payer actually asked us to run a study, which was a pilot study. Now we start collecting since two months now the data. For example, for prostate, we already can imagine a 50% less medical time and a 50% reductions in term of turnaround time on prostatic biopsy. Regarding the inter-observer variability, we also have data. As I mentioned already, this is going to level the playing field because pathology is a kind of subjective type of practice. But for example, one of the big issue with Gleason score is exactly the way the people apply the rules.

Of course, the use of this technology bring people to discuss together the slightly difference that they had, for example, in Gleason 4 versus Gleason 5. The interesting is that now people are going to converge on what is the result that is provided by AI. So, in general, what we say now, we can already say to the payer, look, I mean, in specific fields of pathology where we're using AI, we're already seeing the benefits. But we know there are other algorithm coming, ready to come. Now, we're not using HER2, but HER2 is a breast cancer. HER2 is now used to decide if going to a biologic drug, which is called trastuzumab or pertuzumab or the other more recent drugs. The HER2 story is now very complicated.

It's not only to decide whether it is positive or negative, but because of the new drugs, which is the antibody conjugated trastuzumab deruxtecan, now we need to evaluate what we call the low expression and ultra-low expression. We now presented data at my national society meeting in December, where the use of the Aiforia algorithm is going to improve the quality of HER2 testing, particularly on the very challenging low and ultra-low expression. This is going to change the field of breast cancer prediction when HER2 comes into play. That's the examples I can make to you at the very time.

Moderator

Thank you. Interesting examples. Another question. How is AI in pathology funded in Italy? By the hospital, the region, or national programs?

Angelo Paolo Dei Tos
Professor of Pathology and Dean of School of Medicine, University of Padua

No, at the moment, in Italy, the health services are funded, of course, from the central government, but then the single region decide how to use the money. At the beginning, the first investment came through the funds of PNRR. Actually, in my region, we spent almost EUR 25 million in order to set up the digital framework plus picking up AI. The reason was because we could use the money at that time, in particular for softwares and electronic, not hardware, that is the point. Now we completed the process also with the hardware a few months ago, and we are starting.

The payer is the region, of course, and so there is a negotiation between the network of pathology. I am responsible for that as well. For example, now they decided to go on. The first experience will end formally at December 31st, but we are already extending one year, and we are already running the next tender for the more and more broad use of AI in pathology. I am pretty positive about the reaction of politics and payer on the value of this type of approach.

Moderator

Thank you so much, Professor Dei Tos. Next, we are moving on to Digital and Computational Pathology Market Report landscape and outlook by Katie Maloney. Welcome, Katie.

Katie Maloney
Partner, DeciBio

Great. Thank you. Thank you for having me. I will go ahead and get started. Here is a little bit about me, but this was shared previously, so I will skip, and we can jump right into some of the market numbers, and excited to walk through this with you all. On this first slide here, I do not think this should come too much as a surprise to anyone on the call today, but just wanted to start with a general framing of the discussion, what we will be focusing on. We have the very high-level digital workflow outlined, looking at from slide scanning, slide viewing and image analysis through report generation and image storage. Today, we see a lot of the value that is being realized is on the workflow side of things, faster retrieval, minimal couriers, remote reads.

There's emerging computational pathology tools, as were just mentioned, that are helping identify features not readily detectable by visual review, and also that are increasingly going beyond into emerging tools that are looking at morphological signatures, and kind of must-have tools versus nice-to-have tools as well. To start by looking at the worldwide digital pathology market here, we're looking at about a $1.1 billion market in 2026. You can see that 80% of the market is split between the U.S. and Europe, with about 20% of it in the rest of world. I think a couple of the key points to know about this market is that today, we see about two-thirds of it that comes from biopharma and CROs. There's a lot of spend looking at biomarker discovery, trials, companion diagnostic development, really looking from translational through to clinical trials.

About 1/3 of the market today is on the clinical side of things, so academic medical centers, community hospitals, reference labs, independent pathology labs. Within this market, we have it segmented by product as well, looking at the scanner spend as well as the IMS and AI spend, and it's that software side of things, the software and AI that's really growing faster, or outpacing the hardware spend. This is something where we've seen some of that hardware that's been put in place over the past five to 10 years, and now the use of it is scaling, and the adoption of software and AI is continuing to increase. As far as growth goes, we look at about a 12% growth year-over-year through 2031.

This is driven by a number of different things, which I'll get into throughout the presentation, but we do see quite a lot of room for optimism here within this market. To start by looking at how the market's growing, I think it's important to take a step back from digital pathology into pathology more broadly. For those of you that are familiar with the space, these numbers shouldn't come as too much of a surprise. But within the pathology market, we're looking at more cases that are more complex, that need to be analyzed by fewer pathologists with tighter reimbursement. All of this is leading to a world where the economics of digital pathology make increasing sense. In terms of more cases, we're looking at about six million new cancer cases that are expected to be diagnosed in 2026 between the U.S. and Europe.

There's an aging population that keeps these volumes climbing. The complexity of cases certainly factors in here, so there's increasing number of relevant biomarkers. Dr. Dei Tos was just talking about this previously, looking at things like not just HER2, but HER2-low, HER2-ultra-low, increasing number of relevant biomarkers per indication that increases the average time per case. Pathologists' availability has continued to be an issue within this market. This is not specific to any region. In the U.S., we see that pathologists' workloads are growing and the open positions are increasing. In the U.K., around 80% of pathologists say that staffing is inadequate, and those numbers are reflected across Europe. It's certainly a broad trend there. Reimbursement factors in as well. There's tighter payment, limited availability of funds. Then we can look at digital pathology specifically. What are some of the market dynamics we're seeing?

This is primarily looking at just the past year, past 18 months. How is the market changing? Why are we having this conversation now? Why are we talking about it today in 2026? We are seeing that the hardware market is maturing. I alluded to this already. Scanner performance is converging a bit. We are seeing more standardization of file types, and there is an increasing focus shifting towards the software, service, and integration pieces. The ecosystem does remain open, but consolidation is increasing.

I will touch on this in the subsequent slides as well, but we are seeing a number of M&A deals within the past year that are changing what the ecosystem looks like, but the interoperability of this ecosystem continues to be important. We are seeing regulatory pathways that are expanding, so there is increasing number of clearances, both on the scanner side of things as well as the viewer and AI.

Increasing number of tools that are certified under IVDR are impacting this landscape. Reimbursement is certainly one of the biggest evolving pieces and one of the biggest uncertain questions, particularly when we look at the U.S. In Europe, we see more of a national payer system or regional payer system approach to adoption of these tools. Of course, the role of these tools within precision medicine is growing as well. So, moving from workflow tools that are helping pathologists do things that they were previously doing, increasingly into things like companion diagnostics, prognostic signatures, where these tools become must-have tools in the market.

Looking at the partnership ecosystem, I know this slide is very busy, and if nothing else, I think this visual helps show just how interconnected the digital pathology ecosystem and the digital pathology market is, how there is a number of companies that all have to work with each other in order to make an end-to-end workflow that works within an individual lab.

It combines things such as the scanner, the AI tools, the IMS, the data providers, the LIS providers, potentially the reagent providers, pharma, reference labs, and academic medical centers, and community hospitals all coming together to create a viable ecosystem. As a result of this graphic, it probably comes as no surprise that we are seeing an increasing amount of M&A activity. So we have seen some big deals within the past year. Agilent most recently announced their acquisition of Proscia, Roche, PathAI, Tempus acquiring Paige, Evident and Pramana.

This is not an exhaustive list, but this consolidation is helping to decrease the time to market of these tools and also creating more end-to-end workflows that fit under a single company. We are still seeing a lot of the, on the AI tool and algorithm side of things, there is still openness there, but the ecosystem is maturing upstream. I wanted to touch briefly on a couple key trends within this space. First, looking at reimbursement.

This is going to be slightly U.S.-centric, but I will comment on what this looks like ex-U.S. as well. As I mentioned previously, this is one of the biggest outstanding questions within the market, and so we are looking at what I would still certainly consider a maturing reimbursement market. We see that, although there is kind of one CPT code for computer-assisted quantitative analysis, it is not really widely used for a number of reasons.

There's a handful of Category III codes, both for digitization and most recently for algorithms, but these are generally unreimbursed, even though there's a rate set. One of the kind of most used pathways to date has been PLA codes, which are generally for specific prognostic tests, but these codes aren't being granted since 2024. The question becomes, how are these tools being funded? How are they going to be reimbursed going forward? We do expect more visibility going into 2027. Just in the past week, CMS in the U.S. has proposed this Software as a Medical Service payment category. It would reallocate or move where these AI-specific tools would be counted, which may create new pathways going forward. In Europe, it's certainly a different landscape, as again, Dr. Dei Tos was referring to. This is something that differs country by country.

We typically see it either as a national health system funding or regional health system funding. But we've seen a lot of success with this for a number of reasons, which I'm happy to comment on. But we've seen in places like the U.K., the Nordics, and Netherlands that there's been a success in broad digitization of the pathology labs. We have another slide here looking at the global landscape of digital pathology. This is, again, not exhaustive, and I'm not going to go through this point by point because I know that that's not going to be the most enjoyable for you to all sit through. But I do think it does show that this is not a European-specific or U.S.-specific market trend, that the promise of digital pathology is really a global promise. It allows access to more specialist review, broader access to high-quality care globally.

We're seeing initiatives really across the world to help drive access to these tools from Germany and Italy in Europe, into Japan and Singapore, Australia, in APAC and beyond. This is something that we expect to continue to expand. One data cut that I think is often interesting for people, one of the questions that comes up based on what I just showed is, how digitized are labs? Like, how much access to these tools really exist? As you can see here, this is certainly an evolving question. This is based on some data that we have. We have an internal tool. We do some broad outreach throughout 1,200 labs. We have digitization data, high-level digitization data from about 800 labs, and then granular data from about 400 labs.

That allows us to get to about a 30% number in terms of labs that have access to a scanner. So it's just under one in three, and this is U.S. and Europe specifically. Within that, it's about 25% of labs that have access to a viewer IMS. That means typically that these scanners are being used in one form or another. Then it's about 15% of labs that are using it for routine clinical care, so about half of those labs with a scanner are using it for routine clinical care, the other half using it for things such as research, slide archiving, training, and education. Then it's about 10% of labs that are using digital pathology to sign out cases digitally. So you can see within this, there's really a tale of kind of two growth opportunities here.

There is both significant opportunity for growth within digitized labs to continue to scale the use of these tools, and scale use of image analysis and AI tools. Then there is also growth opportunity in terms of labs that are not digitized today to adopt these tools in the future. We see the upside there, and certainly qualitatively in the conversations that we have on this topic, we see the growth coming in both of those pieces. It is really hard to pull out specifically, is the market going to be driven by one versus the other? Because both hold true as you talk to both users and non-users. Moving on a little bit downstream into the digital pathology algorithm space. This is certainly an active market. It is a diverse market. We see about 50+ active vendors in this space, around 200+ algorithms.

You can see that companies like Aiforia, PathAI, Paige, et cetera, have a good menu of algorithms within this space, spanning different applications and indications. As I mentioned, the regulatory environment is continuing to mature, so we are seeing increasing number of algorithms that are CE-IVD certified under IVDR or increasing number looking towards FDA clearance. We are seeing growth from some of the workflow and QC tools, adoption there increasing and standardizing, and then look towards more complex diagnostic marker quantification, predictive and prognostic tools as well. The use of those tools is certainly heterogeneous today, driven by varied digitization access, varied lab infrastructure, but certainly where a lot of interest lies. There is also increasing interest or focus on explainable AI really across the ecosystem, given some of the regulatory requirements here.

In terms of the types of AI tools, the types of quantitative tools we are seeing on the market, there is certainly some expectations around how this market is going to evolve. This is where a lot of the investment that is made on the hardware side of things, on the upstream integration side of things, sees the value really realized in the AI tools that are adopted.

A lot of times, and I haven't spent too much time talking on this today, but the ROI calculation for a lab can be quite complex until some of these tools are adopted, because in many institutions or many regions, direct payment is limited, as I alluded to. You are looking at things like saving on courier costs and saving on staff retention until you start to introduce some of the AI and image analysis tools that have real impact on the labs.

So things like quantitative marker scoring, this is where we expect to see probably the first AI-based CDx within this market, looking at the TROP2 CDx, expected from AstraZeneca pending trial data readouts. But this is a story beyond that. It is not limited to this single marker, and there is definitely interest both on the clinical side of things, but certainly from pharma and researchers as well into developing these tools, especially any pharma with an ADC pipeline is certainly looking at these tools. There are a number of prognostic and predictive tools entering the clinic. These are tools that are typically looking at H&E stain slides and making assessments around therapy benefit or risk of progression. There were some early FDA clearances here and some expectation for this to continue going forward. Also tools that are looking at other things like molecular markers from H&E.

Oftentimes, these are really more of triage tools today, but it is allowing you to look at what do we think this MSI status is or EGFR status is from an H&E stain slide, which can give a lead time of a couple weeks before NGS or molecular testing is back. There are other things like virtual staining, so looking at predicting IHC or special stains from unstained slide or an H&E stained slide, which has some interest, as well as things like multiplex, spatial, as Dr. Dei Tos referred to, spatial transcriptomics.

Certainly, these are largely research tools today, and likely will remain so in the near future, but there is certainly interest in understanding the role that these can play in identifying patients for specific therapies or drug response going forward. I have two last slides that I will finish up with here before we see if there is any Q&A.

As I walked through, a number of tailwinds within this market that make us optimistic looking forward, that drive that growth story. There is a need for it, just an inherent need for it from the pathologist side of things, driven by the workforce gap, more complex and rising case volumes, biopharma, and biotech, CROs continuing to invest in this and continue to have interest growing in this, and this is really something that we see across the space. The regulatory and reimbursement pathways continue to mature, and there is still that large untapped base. Within some of the key opportunity areas, things such as workflow orchestration continue to remain important, AI-based CDx and H&E tests, increasing role of image analysis tools within this ecosystem. Community hospital access is certainly something that is of interest.

This hub and spoke model is increasingly prevalent and driving interest in these tools, and then the maturing data infrastructure. Some of the things that we are keeping an eye on is that clinical uptake is still relatively early, so there is a lot of growth potential there, but keeping an eye on that story, looking at the evolving reimbursement landscape, as well as the fragmented AI market, and the distribution of how these tools are accessed. Just finally, before I end it, to take one more look at the market here, we can see that growth that we have projected. In terms of what the moderators are, I have already touched on a lot of those things such as questions around reimbursement, and the regulatory landscape that is still evolving.

In terms of the drivers, it is really clear what is driving this market growth through 2031, things such as the importance of open platforms at scale, these platforms that are increasingly scaling, standalone regulatory clearances, the high complexity AI, and emerging image analysis and CDx tools, and then really that multimodal data integration. I will pause here, see if there are any questions, but I thank you all for your time.

Moderator

Thank you for these insights. We have a few questions for you.

Katie Maloney
Partner, DeciBio

Okay.

Moderator

Firstly, forecasts for AI pathology are large, but the revenue the industry actually earns is still modest. What is holding it back, and what do you think will be the turning point?

Katie Maloney
Partner, DeciBio

Yeah, absolutely. I think that it looks a little bit different across different segments of the market, certainly. This is the global market that we are looking at. We are seeing real spend from biopharma, biotech today, looking at things such as algorithm development, just digitizing slides so that they have access to all that data. I think there is increasing recognition of the value of the data that is being generated. On the clinical side of things, there is certainly still room for growth there. I think that part of that story is that the reimbursement and the actual payment mechanisms are still maturing. As long as the equation is not as simple as do test X, get paid for test X, the decision is very lab-specific, it is very organization or institution-specific in terms of what justifies or makes adoption make sense within that institution.

But I think as we see the tools such as those from Aiforia, but on the AI side of things continue to mature, that increases the clarity of the value proposition as well as maturing reimbursement landscape, and those are certainly tied together. It is a little bit of a feedback cycle there. We will see the revenue continue to grow in turn on the clinical side of things.

Moderator

Thank you. How are the Roche PathAI and Agilent Proscia deals changing the landscape for independent AI vendors, and does this create a distribution opportunity or does it squeeze them?

Katie Maloney
Partner, DeciBio

It's a good question, and I'll say it's certainly still early with both of those deals. I think we're going to see how that market is going to evolve as we see how what's happening at the top trickles down to how individual labs adopt these tools. I would expect that it's going to create an opportunity in the near term because oftentimes for labs that maybe don't have a clear champion of digital pathology, the adoption has been complicated to date. You have to decide, what scanner do I want? There's a bunch of vendors here. You have to decide, what IMS do I want? There's a bunch of vendors here. That's decisions you have to make before you even get to the algorithm side of things.

Now, with companies like Roche and Agilent that already have relationships with these labs, building some of that end-to-end workflow themselves internally through the image management, they will be increasing access to the AI tools. Most of these companies do some type of AI store type model today, where a lot of algorithms are available on their platforms, and you can see from the interconnected diagram, many of the AI vendors have partnerships with these companies. I think it'll help kind of the majority of labs where there's interest in these tools, but the easier the upstream adoption is, the more likely they are to have access to downstream tools, will certainly be like a group that will benefit. It's also some infrastructure that's necessary as we talk about things like CDx, where you have a single medical device that needs to be available.

Jukka Tapaninen
CEO, Aiforia Technologies

All right. Thank you, Katie. Let's move forward. Obviously, we have some time for questions at the end of the presentation as well. Maybe I continue from there from Aiforia's perspective. This is the market that Katie just explained. The kind of how is it going, what type of elements there. We are talking about the image management system, scanning. Then, of course, as Professor Dei Tos was referring, that, okay, the only reason to do the digitization is to do the diagnosis and use the AI. The AI is then kind of enabling the prognostic and predictive type of things happen. But the market is already sizable and it's growing. On the last question about the Proscia or PathAI, nowadays I guess we should talk about Roche and Agilent.

I can say that, we have a strong partnership in place with those companies, and I have been talking about the consolidation before, and it seems to be happening now on an image management level. I assume that next step is the diagnostic part and the AI part of this workflow. We are working very closely with both companies, and of course, there are other players, strong players in the market as well. We work closely with Sectra, a Swedish company. We work with Techcyte, the U.S. company, and as you know, Techcyte actually now is the image management system for Mayo Clinic. We are having already an integration with them, so I am working closely with. But they are PathPresenter and a couple of other major players in this industry.

From a consolidation point of view, it is really interesting. So that, who is leading the image management side, who is leading the AI side? We feel that we have currently the strongest offering in AI tools. There is a certain time of the market, and we have to acknowledge that the customers, they need to invest on a digital pathology. They need to have the scanners, they need to have the image management, and the next logical step is the use of AI. But it is a step by step. In some cases, we have been able to go with the AI lead first before they have a large volume of scanners. But I would say that that is more an exception of the thing.

When we talk with the customers, often comes up, the ROI on all of these investments. The actual ROI story is fairly simple to do from an AI perspective. If you think about the AI-assisted diagnostics, we can reduce the time, so we can make the pathologist much, much more efficient. There was this lack of resources, but then we provided tools that you can actually automate certain parts of the work and enable new parts of the work, and that actually brings in the ROI story. The other thing is that we can obviously also reduce some of the more expensive stainings that are needed for diagnosis. When you use the AI, you get a lot of savings on that side as well.

All in all, the investment that you have already made on the scanners and image management, it is actually then give the return when you start using the AI, and that is the logical next step to happen. What comes to our situation in the market, we have been, as you know, so we have this large community of researchers using Aiforia Create, which is the development platform. That is where Aiforia started back in 10 years ago. Those tools that you can build your own AI models, and there was a lot of scientific proof coming out of that exercise. That is actually the core asset that we have in the company when we develop these AI algorithms, AI-assisted diagnostic tools for clinical use.

We started to do that exercise about four, five years ago, so when we listed in Helsinki and get the funding. We started to build a portfolio of AI tools for clinical use and getting all the regulatory parts also sorted out. On that roadmap, we have been moving forward really nicely. We have the portfolio, which is the largest at the moment. We have 11 CE-IVD, IVDR marked solutions in the market. We have the regulatory approvals that we can sell on a clinical, which is not so common. As Katie was mentioning, there are 200+ companies doing some algorithms, but very few of those can actually sell on a clinical side. You need to have all the kind of science-related regulations in place, and then AI-related and technology-related regulations in place.

We have invested a lot of time and effort on having those done. We have a really competitive portfolio in place. But then the customers, when we started to go to a clinical side, the idea was to get the leading hospitals in the world to customer in the first place, and then be as a reference. We made some nice success with the Mayo Clinic, with NHS, and Paris Hospital region. Of course, the Veneto region was one of the great kind of customers for us, and still is, Lombardia region and so on. We have been kind of getting those contract signs, and we also have signed some additional things in the recent months as well. Of course, hunting new ones.

I just wanted to point out that this was maybe a kind of a small surprise how long time it will take to getting into a full volume. My point on this slide is really that, okay, so we have already + 20 clinical customers in place. Seven of those have been kind of delivered, and it is in a productive use, not in a full use yet, but it is already in a kind of use with the pathologists in a routine, and ready to expand. Then we have some customers that we have delivered, but they have not yet started to use. Then we have a bunch of customers that we have signed the contract, and we are in a delivery phase, and the usage will start when it is delivered.

There has been some slowing down factors that the customers did not have enough scanners, or they did not have the LIS system, which is a laboratory information system implemented, or they were changing that, or they are kind of changing the image management systems. But once those things are sorted out, then this starts to be already a really nice group of customers that will turn to a kind of steady revenue flow to Aiforia as well. But this is something that I would say none of the competitors in the market has not been able to achieve, so that you have signed contracts, you have delivered the solution, and it is already in a productive use. Two sectors I mentioned already, the clinical side and the research side.

They go together because in many cases, the hospitals and clinics and lab chains, they want to have the research tools as well. They build their own AI models, and we can provide them a roadmap to a clinical use as well. A good example here is the Mayo Clinic, where we have the breast cancer suite on a production, but they already have 70 pathologists working with the Aiforia Create tool to build their own AI models for research purposes, and then also bringing some of those solutions on use on a clinical side. It is a good combination, and I think it is a unique one as well in this industry, that it is not a black box thing.

Our really competitive advantage is that we can show how the model is built, and we can also provide the tools for researchers to build their own models. Not going on too much on the details, but this gives the credibility on clinical solutions as well. There are thousands of AI models already done with our technology. It is really tested very well, and it is proved that it actually works. This is something that a lot of research is leveraging already today. That gives us a good point on building these clinical models as well. We start first with the biggest volume area, we look at the prostate. Prostate is in high demand, the breast cancer suite is also an interesting one.

Then we have a colon and gastric and other models coming up. Thinking from our side, when we started to do this was that when you go and start automating, you need to have a full portfolio of tools. Not just a one-point solution, but you bring in multiple tools that you can provide the same user experience, user feeling, and same integration and maintenance cycles for these tools.

That also the user feel comfortable getting the benefit on using the tool, but then the big organizations, IT department, and a C-suite is feeling comfortable that it actually brings the return on their investment, but also that IT feels comfortable that the maintenance and integration is doing fine, that this is a real software that they can implement on their systems in a hospital. Then of course the benefits. There was a lot of discussion from Professor Dei Tos about the predictive and prognostic models. We have actually done some of that already with the Mayo Clinic, so we have one solution called QuantCRC, is for colorectal cancer, where you can predict the patient outcome, and based on that information, you can decide how much chemotherapy the patient is getting.

But all in all, what we are providing today is a faster care path. You get, first of all, in most of the places, you have a lack of resources. We provide the tools that you can actually do the work, but you can be more efficient, and also the quality is improving, so that you make less mistakes. Then by doing these things, you can also provide an access to a kind of a high-quality diagnosis in remote locations as well. The equal access on a high-quality diagnosis is available. This is of course enabling then the kind of predictive and prognostic models as well. Why Aiforia? We have the technology leadership. We have been in this industry since 2014 when we started.

We started actually on an image management side, so that you can store and share and visualize the images. But in the early days, we figured it out that most of the value comes that you actually understand what is on that slide, and we started to build the AI tools. We have been there with the kind of technologies at that point of a time, convolutional neural networks. Lately, we have been moving to kind of the latest technologies, enabling the models with Foundation Engine and foundation models, so that we have been further developing those things. But we have that technology leadership, and that is why we have quite a heavy investment on the R&D side as well. Then the scalability.

This is a software company. We have been able to kind of build now the models in a level so that it starts to be almost like an out of the box that you can deliver and you can scale up. The delivery time from our side has been shortened significantly, and also the generalizability of the solution has been improving big time. Now if we talk about the breast cancer or prostate cancer, we can just take the customer slides and we can run it through the system. We can already give a good results to use. Of course, one of the benefit what we have as a company has been always that we can secure that the results are absolutely top level, because we can always train and calibrate a little bit with the customer slides.

Then because the stainings or scanning gives a little bit different image quality, then it is good to have that opportunity to kind of calibrate a little bit. But the business is scalable, and the demand is high, and now we have the tools in place so that we can capture the opportunity. Of course, the interoperability certified and certification and security, it is really, really important as well, because you are not alone. You need to be able to kind of support multiple scanners, multiple image management systems, multiple LIS systems, so that you need to be able to kind of operate in an environment that has multiple different software and hardware kind of companies in play. We have done all of those things already with all the major kind of players in this industry.

The certification and security certified quality, those type of things, it is absolutely a must, because sometimes there are misunderstandings that, okay, I build an algorithm so I can bring it to a clinical use, but unfortunately, it is not so simple, so that you need to have these EU, U.S. technology level kind of certifications and regulatory things in place so that you can actually sell on that industry. It is also supporting us a little bit, but also it is a kind of one of the areas that we are investing a significant amount of money and resources to make it happen. That is from my side, and now there is time for questions. Thank you.

Moderator

Thank you, Jukka. First question. You have signed significant contracts, including AP-HP in Paris and healthcare regions in Spain and Italy. How long does it typically take for a signed contract to turn into revenue?

Jukka Tapaninen
CEO, Aiforia Technologies

That is a, I would say, very good question, and it varies. We have some cases that it has happened in a month or two. We have some cases that we have sold two years ago, and it is not yet turning up or usage because of multiple reasons. I would say that, okay, typically, we are talking about half a year to nine months time when it starts to be a revenue. I just wanted to point out so that it is not that we couldn't deliver the Aiforia part, but it is that, okay, so the customer is typically doing a big transformation at the same time, so meaning the scanner acquisition, maybe the LIS kind of a change, so kind of implementation or image management system implementation.

That is the combination that you have to have in place, and then you can start using it, and then it starts to turn as a revenue for us as well. Even it takes some time, but we believe that, okay, so it is a sticky business that once you have that system up and running, it is likely that you continue to use and increase the usage. We have also a couple of customers who have been buying and then implementing and using, and they are actually using much more than originally planned.

One really important metric for us currently is that, okay, so we look at those +20 customers what we have and how the things are delivered, and is it in a productive use already, and then how the productive use is actually growing. On that side, we have really good numbers from, let's say, beginning of this year. We have been more than doubling the usage of the software. That just proves that the product works, and once it is in use, then you want to use it more.

Moderator

Continuing from that, what does your target of 50 key accounts, each with EUR +500,000 in annual recurring billing potential look like in practice? What does a customer's path to that level look like?

Jukka Tapaninen
CEO, Aiforia Technologies

Well, yes, that comes from our targets, and I think it is one of the key targets what we have. As I said, we have currently + 20, and we have a nice roadmap moving forward and plan to have that 50 in a place or even more than 50. The idea obviously on target was that, okay, we are targeting on organizations that has a possibility to go over than EUR 500,000 a year level. It does not happen in year one or year two, but okay, eventually those customers what we have in a play at the moment, they all should be able to use the software in a level so that gives us significant revenues as well.

Moderator

Thank you. The final, well, maybe not final actually, but next question is that how is the commercial cooperation with Siemens Healthineers progressing?

Jukka Tapaninen
CEO, Aiforia Technologies

Well, Siemens, it is a large organization. They have had their internal changes, so now it is, I think it is now called Healthineers. It used to be Siemens Healthineers. It is a good level for collaboration, so we have been building a pipeline. We have been training the Siemens sales force. As you know, we were selected as the AI vendor on their offering, so there was one image management vendor, that is Proscia, nowadays Agilent. Then of course the Siemens own hardware and software solutions top of that. It is moving forward, but it has been, I would say that, frankly speaking, I would have expected a bit faster start on that. The collaboration is good, pipeline is building up, and we expect to get some deals signed with them as well and moving forward with that.

Moderator

Thank you. Next question. When do you expect Aiforia to become profitable? Do you need to raise new equity before your cash flow turns positive?

Jukka Tapaninen
CEO, Aiforia Technologies

I have to say that also very good question, but I cannot answer all of those questions at the moment. We have a plan, of course, now significantly increase the revenue levels, and we are having strict cost controls in place as well. We are moving that direction at the moment. Whether we are doing an additional funding rounds or not, that has to be seen.

Currently, we are in a good position, so we have this European Investment Bank loan agreement signed, so that is potentially covering a EUR 20 million funding. It is in three tranches, and that is a good starting point. I would not like to say yes or no whether we do an additional one, because if the market starts picking up faster and nicely, it might make sense, but I cannot say that today. Currently, funding is secured, and the situation we have a + 12 months runway in place, and we have the EIB contract signed up.

Moderator

Thank you, Jukka. That was our final question. Thank you also for all the guest speakers and our participants.

Jukka Tapaninen
CEO, Aiforia Technologies

Thank you so much.