Reti Health: A Year In Review
On December 11th, Reti Health will be one year old. Given such a glorious anniversary and to celebrate that the company is on the verge of finally hitting the market, I thought I’d provide Reti Health’s 2021 Wrapped.
My entrepreneurial journey started in September 2020 when I joined Entrepreneur First’s latest London cohort. Entrepreneur First, or EF as it’s known in the lingo, is a talent accelerator. In practice, this consists of finding 80 bright wannabe entrepreneurs and lumping them together in a room to pair and break up with each other until they find a suitable co-founder. If this sounds like Love Island to you, your instincts are correct. It is basically the same thing, with far less sexy nudity and far more sexy discussion about NLP.
Despite being certain I wanted to start an AI MedTech business, I found myself teaming up with a host of different characters to try and validate a series of disparate ideas. First, I researched the potential for chatbots to help mental health patients. Then I spent a week looking at building an open banking payment tool that would let you pay with just your face. Finally, I spent two weeks calling telco operators worldwide to understand if GANs could be used to provide video calling in areas with poor connectivity. Needless to say, each idea was a bit of a non-starter.
To cut a long story short, I found myself entering the last few weeks of the program without a partner or an idea. I’d pretty much given up on the premise that EF was going to be my entrepreneurial launchpad. I’d started applying for jobs and had even set up a few interviews. However, I was still going through the motions of talking to other “single” cohort members. My experiences in the program had crystallised one thing for me; I knew I wanted my next career move to be in MedTech or BioTech. Consequently, I was pitching a series of undercooked medical-related AI ideas to all I spoke with.
On November 18th, I chatted with George. This wasn’t the first time we had spoken. However, we had evidently never clicked before. We will never know whether this was because he felt intimidated by my boyish good looks or he was jealous of my extensive vocabulary. Regardless, during this conversation, we agreed to explore the idea of using AI to diagnose optical diseases more accurately.
We quickly arranged market research calls with two optometrists for the next day. These calls were illuminating. Firstly, we concluded that there is likely no commercial market for AI to diagnose optical diseases in the UK better. Secondly, we concluded that there might be a commercial market to use AI to assess retinal images to provide risk stratification for systemic diseases, such as cardiovascular disease.
These discussions encouraged us to form an official team to validate the idea further. We devoured relevant academic papers, reached out to leading researchers, hustled to get a call with the Specsavers’ founders, and spent a day cold calling optometrists. Additionally, we attended eye exams at our local optometry stores resulting in me picking up such a bad case of conjunctivitis that I ended up in the eye hospital.
On January 18th 2021, George and I stepped in front of EF’s Investment Committee. Despite weaving (what I imagine was) an inane and incomprehensible story about how Reti Health would one day become a unicorn, EF and some generous angels decided to seed us $130,000. Suddenly, the company had got very real, and we had investors who expected us to build something massive.
Next was EF’s demo day in March. A demo day allows startups to present their products and services to a specially selected audience of investors. EF’s demo day is fantastic. It provides exposure to over a hundred leading European early-stage VC funds and angels. Unfortunately, we still hadn’t managed to secure the critical ingredient to our product: data. Specifically, tens of thousands of retinal images linked to cardiovascular disease outcomes. As a result, we were still raising off a deck and unsurprisingly got short shrift from investors.
This marked the start of the “dark period” characterised by a desperate and frantic search for data. We talked with leading research hubs focused on ocular medicine. None had the type of dataset we required. Two informed us that they were building such datasets; however, both came with the caveat that all the patients would have diabetes and they did not expect to finish gathering them anytime soon.
Undaunted, we pursued a series of harebrained schemes to try and build our own dataset. First, we attempted to recruit members of the public who had previously undergone eye exams and have them make a GDPR request to their optometrists to get their retinal photos. Unfortunately, the GDPR requests proved too time consuming to ever be seriously scalable. Next, we bought a handheld fundus camera and set about imaging our friends and family. Sadly, acquiring good images with a handheld fundus camera is exceedingly tricky. As such, we also jettisoned this approach.
The lowest point came towards the end of summer. We convinced the Hakim Group, the fourth largest optometry group in the UK, to make a joint application for an Innovate UK grant. We heard from a series of grant consultants and advisors experienced with Innovate UK that we would likely have a very competitive bid. Consequently, we were despondent when we found out that not only had our bid been unsuccessful, but our score was not even competitive.
Happily, the “dark period” came to an end. In October, we got our dataset. 175,000 retinal images across 85,000 patients linked to complete systemic disease outcomes, including cardiovascular disease, neurogenerative diseases, anaemia, and diabetes. Anxiety that we would not be able to train working models quickly subsumed our sense of relief.
Fortunately, our concerns were misplaced. By the end of November, our models were achieving the same level of accuracy as the benchmark, non-retina based cardiovascular disease screening algorithm (QRISK3). This has allowed us to progress to plugging our model into our web app and certifying the entire product as a medical device. A pilot study will be the final stage before a full commercial launch. We are currently recruiting optometry stores to take part during Q1 2022.
To summarise, it has been a wild first year at Reti Health. There have been some crazy highs and crazy lows. We’ve nurtured a small dedicated team, secured access to a fantastic dataset, and are poised to determine if the market actually wants what we are selling!
We go into the new year dizzy with excitement like a toddler drunk on lemonade. In the next few months, we will find out if we have invented a sweeping new technology that will help stop many preventable heart attacks and strokes.
A huge thank you must go to my co-founder, George, for putting up with my inane babbling, Eddie Korot for his domain expertise, Imran Hakim and the Hakim Group for his/their patience and thoughtful guidance, and our fuzzy friends Clover and Sukki.
Are you Reti for the future?
Our website: www.retihealth.com
