CEO PERSPECTIVE: Jay Lakhani, CEO of Visulytix


    Your name

    Your email address

    Your message


    News & Events


    Visulytix is using Deep Learning based Artificial Intelligence in Ophthalmology to deliver clinical decision support solutions which permit the early detection of sight threatening conditions. Ultimately, prompt & accurate detection results in the best possible outcomes, benefiting patients, providers and payers, whilst supporting the transition towards value-based healthcare.

    Visulytix’ Pegasus platform is an AI-powered software that an optometrist or ophthalmologist can access to help them analyse retinal imagery. It searches for features that are indicative of:

    • Diabetic Retinopathy
    • ONH pathologies including Glaucoma
    • Wet or Dry Age-Related Macular Degeneration
    • Diabetic Macular Edema
    • General Macular & Retinal Anomalies

    Coulter:Pulse recently interviewed Jay Lakhani, CEO of Visulytix to find out more…

    1. Tell us about the role that Visulytix is playing in the world of AI? What do you see as your key opportunities?

    Jay: Around 700 million people worldwide are affected by the three major blinding eye diseases that we are targeting, namely glaucoma, macular degeneration and diabetic retinopathy. Diabetes is developing into a global crisis, while the increased incidence of age-related conditions like glaucoma and macular degeneration reflects an aging worldwide demographic. Around 9% of people above the age of 45 have macular degeneration but a large proportion remain undiagnosed until they have lost a significant amount of their sight. Because the brain is so adept at papering over the cracks and masking faults, we are slow to notice the signs ourselves.

    A regular eye check is critical to enable the early diagnosis of vision problems and identify changes that may indicate future conditions. Only once an accurate diagnosis has been made can the eye care professional provide a detailed explanation of the condition, as well as the potential prognosis.

    We started building a massive heterogeneous dataset years ago to ensure the highest degree of accuracy for our machine learning models. We have now gathered data from across the world, trained the algorithms and created a product. An optometrist who takes a photo of the back of the eye can upload that photo in the form of a JPEG or a TIFF file straight from the camera onto our Pegasus platform, where it is then analysed.

    Crucially we are not diagnosing but assisting the diagnosis that is being made by the optometrist. Via Deep Learning the algorithm is trained to look at images of the back of the eye and infer if something is wrong. We are not comparing images to a normative database but taking core content from images and saying certain content suggests a certain disease. The eye has the only visible nerve ending in the whole body and a functioning optic nerve can be viewed directly without being affected, unlike other nerve endings in the body which can be accessed only via some intervention. A whole range of factors can be identified on the fundus, for instance beta or tau amyloids, which you would expect to see in Alzheimer’s patients. The eye has the smallest blood vessels in the body and anything that affects blood vessels can be detected, for example signs that may subsequently suggest a risk of stroke or cardiac disease. Other macula conditions can also be identified such as macular degeneration.

    1. What do you see as the most promising Artificial Intelligence advances in the field of Ophthalmology?

    The biggest advances in AI are in deep learning, and the opportunities for ophthalmology are greatest in the imaging space by supporting the early detection of blinding eye diseases. AI in platforms such as ours are helping to improve overall diagnostic capability, and in doing so permit skill enhancement. Furthermore, the result of improved accuracy is efficiency. Such systems can help reduce the number of false negatives, where a patient is incorrectly diagnosed as healthy, or the number of false positives, where a patient is misdiagnosed as having a condition. Patients want the very best care, payors want to pay as little as possible and doctors are overwhelmed. And patients are of course 100% on board with getting an early diagnosis, even though the news may scare them initially. All 3 conditions I mentioned have genetic predispositions, and are particularly serious in some population groups. There are also coupled risks if associated with myopia, which is also on the increase in many populations, as a result of smartphone and technology use.

    1. What challenges do you face in this market? For instance, from regulatory bodies or from the competition?

    It is critical that any medical device with an effect on patient outcome is regulated. As a young company, however, this is a huge burden for us financially and in terms of human resource. We rigorously follow our processes to ensure we are compliant! These circumstances of course favour companies like Google or Amazon who have deeper pockets but it is absolutely necessary and correct that every company affecting patients are held to a high standard. Patient safety must come first.

    Among the competition, IDx has recently gained FDA approval for their AI-based diagnostic system to detect diabetic retinopathy. For macular degeneration, a different type of 3D rather than 2D imaging is supplied from an OCT device.  Our technology can use an OCT input rather than just a fundus. Our main competitor here is Google – you may have heard of them?! Our rationale was “why only screen for one condition when you could screen for the others as well?” As a data science puzzle, this is an enticing challenge and of course a great differentiator for our business. Our next step will be to scale up the business.

    1. What sort of talent implications do you see ahead in driving your business towards its goals? Data Science seems to be at the heart of what you do, but how critical is the Life Science element?

    Our biggest challenge is finding good data scientists. We have found that there are 5 or 6 universities producing excellent talent, with London’s universities being very strong. We are more focused on people with machine vision expertise rather than the other disciplines such as natural language processing, i.e. people who can get a computer to understand what it’s seeing. Another challenge for us is that many of our staff are EU citizens and so we are very vulnerable in the face of Brexit. Unfortunately, we can get no guidance on this. It would be a catastrophe for us to lose many good people and this is a terrible disincentive for early start-ups building a company in this country. Another area of concern is cashflow – data scientists are inevitably in high demand as a result salary demands are increasing, even for new graduates without a PhD.

    Perhaps the data scientists of tomorrow will come from China or India?

    It will be interesting to see how Europe deals with this challenge! I have no doubt that both India and China are becoming leaders in this space in their own rights.

    What about talent shortages at the more senior level? 

    At the senior level, the competition for talent is driving remuneration sky high. I’m not sure the right candidate is necessarily going to be sold on the start-up culture or the promise of equity alone.

    Nevertheless, we are well connected and proactive. The recent meeting at the Crick Institute with Theresa May and Indian Prime Minister, Narendra Modi has created a lot of interest in us.  I recently went on a trade mission to India with Vishal Virani from Ada Health to sell UK health innovation over there, to a very warm welcome. (Ada Health interview can be found in Coulter:Pulse – https://www.coulterpartners.com/digital-perspective-ada-healths-dr-vishaal-virani-on-the-ai-symptom-checker-app/) We also continue to talk to graduates at UK universities and are identifying a future pipeline in the Data Science area.

    1. How did you end up in this innovative “AI in Life Science” space?

    It was not an obvious route! I was working in investment banking in real estate, but had conducted a project in MRI stroke imaging as an engineering student at Oxford. It occurred to me that Data Analytics would soon be as big in medicine as it was proving to be in finance, once the regulators were on board. Then, by chance, I met an Ophthalmologist, who wanted to set up an AI business to meet a need for screening serious eye diseases. I quit my job the next day and started raising money with just a PowerPoint deck.

    We hired ourselves a CTO who really understood deep learning and he took a very big risk joining a company which was effectively unfunded. Then within a few months we received funding from Angel investors based on his early stage proof-of-concept. Now Visulytix is going strong with 16 people and three board advisors who are investors within life sciences.

    Further information can be found at http://www.visulytix.com/