Video Transcription
Andrew Kadis 00:02
Andrew, Hi, I'm Andrew. I'm an engineer. I have got 15 years of experience with complex devices, including the world's first Martin hailer. I'm the CEO of Cambridge Vision Technologies, and we are developing an early detection Alzheimer's disease test to be used at the optometrist. I'll get on to in a minute why the optometrist is a crucial touchpoint. But first, just a bit of a reminder about Alzheimer's disease. One in three people do not even realize they have it. The people in this room, one in two will be directly impacted through an immediate family member. Economically, the burden is huge. By 2031, a trillion will be spent annually in the US alone on Alzheimer's disease care. But right now is a time of great excitement in neurology. The traditional narrative is that once you are diagnosed, no interventions are possible. That is changing. Last year, two new therapies were released onto the market, lecanemab and donanemab. These two monoclonal antibody-based therapies slow down cognitive decline, and they work. In addition, this year, we've seen the release of a number of high-profile, highly accurate blood biomarker tests, and this has been supported by a great deal of innovation over the last five years in digital biomarkers and cognitive-based assessments. So what's going on? What is being built is a large infrastructure of diagnosis, machinery, and treatment for Alzheimer's disease. But there's a problem. We call this the capture problem. If you are someone who knows who has Alzheimer's disease, family history, if you're well educated, then you know to go and seek out clinical help. However, there is a large portion of the population who will not do this; they will wait until symptoms are advanced, and then they will seek clinical help. These drugs are disease-modifying therapies. They slow down progression. They do not stop it. They need to be given early. This is the capture problem, and we struggle with it in the UK, where we're based, but it's also a major issue in the US, where there are well-known neurological deserts. So what is our solution? Our solution is to target the optometrist. As we age, our eyes lose their acuity; we all visit the optometrist. 86% of people 55 and older visit the optometrist every two years. And it just so happens this is the absolute exact same demographic who need to be screened for Alzheimer's disease. Hence, we are using the optometrist as the infrastructure by which screening can take place. Our product is an imaging-based AI software biomarker platform. We use the latest posterior imaging technologies—optical coherence tomography, scanning laser ophthalmoscopes, and fundus imaging—and these technologies have the capability to image the eye, which is effectively part of the brain, at 200 times the resolution of the highest accuracy MRI machines. And why can we do this right now? Well, three things have happened. Firstly, I mentioned the therapies. There are two of them on the market, but there's a very strong pipeline. So right now, there are 132 therapies being trialed, many promising candidates in phase two and phase three, and we envision that these will come onto the market in the next few years. Additionally, there have been major advances in machine learning. So the very first FDA approval for a fully autonomous AI screening system was actually given in ophthalmology digital diagnostics in 2018 for diabetic retinopathy screening. Not only that, there have been significant technological advancements. So what we can do with vision transformers, mask autoencoders, and also foundation models lets us pick out signals that we couldn't discern five years ago. And finally, there's the imaging hardware that sells. It's just getting really, really good. There's no other way to put it. I mentioned earlier that we can resolve a finer resolution than the best MRI scanners, and basically, the clinicians don't really know what to do with it. It's the perfect problem for AI, and this hardware is already deployed at your local optometrist. So our technology, so I won't go too much into the details, but basically, we use a foundation model-based architecture. So this is a dual-stage architecture, and the elegance of it is that it lets us train on both unlabeled and also labeled data. So we take an OCT image and a fundus image as an input, and we produce a disease category and a risk rating as an output, sizing the market. So we believe that by solving the capture problem, this will expand the current Alzheimer's diagnostics market by 5 billion. Looking at a 15% take rate and then a 20% market penetration, you're looking at 150 billion. The other long-term trend is that if you detect more people and you detect them earlier, it's going to lead to significantly larger therapy revenues. By 2032, the Alzheimer's disease therapy market will be 25 billion. So we have a phenomenal team. We're based in Cambridge, the UK, and we've done incredibly well out of recruiting directly out of the university and also the local biotech cluster, which is one of the best biotech clusters in Europe. So we've got plenty of expertise in AI, machine learning, data science, and imaging. In terms of the ask, so we are currently wrapping up our government contract, which is currently funding us. First of December, we're looking at opening a seed round of 5.8 million, and that will allow us to hit three key milestones. Firstly, we're going to take our current algorithm and augment it with significantly more data to a production-grade algorithm. So that's what we're going to take through FDA approvals. Secondly, we're going to kick off a multi-site clinical study as a follow-up to the work that's been done so far. And then finally, we're currently in discussion with commercial partners. So this will enable the commercial pilot. If this is something you're interested in, please reach out and become part of our journey. You.
