Many months after the last Start Up of the Month, I caught up with engineer and founder Isabella Degen to find out about an exciting new machine learning start up for people with diabetes: Ayuda Heuristics.
What is Ayuda Heuristics and how did it arise?
I’m a robotics engineer, and I’ve had type 1 diabetes for over 22 years now. The idea for Ayuda Heuristics started forming over a decade ago when I was doing my masters at the ETH in Zürich. I was fascinated by anything to do with computers behaving intelligently: machine learning, bio-inspired algorithms, robots. I’ve always felt that robots are there to do things that we don’t want to do, and I dreamed of offloading my diabetes self-management to one. About a year and a half ago, I joined forces with Cyndi Mitchell, a technology business leader who was also my mentor, to do just that.
At Ayuda Heuristics, we’re bringing science, people and engineering together to overcome fundamental gaps in the scientific understanding of diabetes. Because diabetes is not understood, people are filling the gaps with their own daily self-care heuristics. Along the way, they’re creating new knowledge about diabetes that they hold in their heads. This a huge cognitive and psychological burden, and even with all the work, it’s next to impossible to achieve the clinical targets.
We’re making a mobile app that learns and takes on each individual’s heuristics and lightens their load. With it, we’re bringing together the heuristics of millions of people to formalise this previously untapped knowledge base and expand understanding.
Why hasn’t anyone else done this before?
We’re approaching diabetes from a very different angle. We don’t think diabetes is a disease: we think it’s an umbrella term for the symptom of high blood glucose. Under this umbrella, there are crude classifications like Type 1, Type 2 and MODY, but other than that, very little is known about diabetes.
Because of the lack of knowledge, there’s only a very rudimentary treatment of drugs, diet and exercise, and people are left to develop their own heuristics to map that treatment onto their own condition. To make an app that really helps them with this, we have to venture beyond the current medical understanding and into the unknown. It’s a big challenge, but it’s a perfect job for a robot.
What is your business model?
We will sell a monthly subscription to our app directly to the consumer via the app store.
There’s an activated world of people and organisations who are already pushing for better diabetes technology. Social movements of people with diabetes and their families are hacking their existing devices and building new technology to make their lives better. Activated peer communities are driving peer-to-peer support without the medical establishment. Diabetes charities and advocates are pushing to accelerate access to technology for people with diabetes. We’ll tap into this world and join the conversation, and that’s how people will find out about us, and hopefully decide to give us a try.
Once they become users, the more they use our app, the more it will learn about them. It will give them increasingly better and more personalised guidance everyday, and take on more and more of their self-care work.
All the data our app collects will be anonymised and placed in a data cooperative that is owned and governed by our end users. We’ll still have some rights to use it, but the cooperative will be able to commercialise it as well, so there’s potential for economic return for all of our users too.
Have you had any success stories yet?
We have hit a lot of big internal milestones so far, but probably the biggest highlight has been the feedback from the people with diabetes who have helped us shape our product. There’s nothing better than sharing your prototype with a user and hearing them say: “This is really useful. It feels like you’re right there alongside me, and we’re going to do this together.”
We’ve had a lot of moments like that over the past year and that’s what keeps us going.
Change can be hard to bring about in healthcare, what is your strategy?
The only way to give people something sophisticated enough to learn and take on their heuristics is to make a regulated mobile medical app. That means CE Marking in Europe, and FDA clearance in the US, so a big part of our strategy is to work iteratively in collaboration with the regulators. We’ll release our product incrementally, proving its safety and efficacy over time.
We think this is a great opportunity to break new ground for all the direct-to-consumer regulated mobile medical app makers out there. We’ll never be able to exploit the full possibilities of mobile in medicine without the regulatory bodies. We all have to work together and help each other.
What are you looking for right now?
We’ve validated our product concept and mapped out our regulatory strategy, and we’ve just kicked off our process to raise our first round of external funding. We’re looking to meet visionary investors who share our thirst for solving complex and meaningful problems.
What is your goal for Ayuda Heuristics in the next 10 years?
We have an incredible opportunity to bring science, engineering and people together to make a major dent in a major world health problem. The science around diabetes is full of gaps and people are filling the gaps with their own heuristics. This a huge cognitive and psychological burden. We use engineering to learn their heuristics and offload them to their phone.
With millions of people using our app, we’ll bring together and formalise all their heuristics. This will be a massive source of previously untapped diabetes knowledge. We’ll be able to identify clusters of people with similar heuristics, and help them borrow heuristics from each other. We’ll also share this new knowledge with the entire diabetes ecosystem: life sciences companies, research institutions, healthcare providers, and the governments that are drowning in the current sea of unknowns.
Disclaimer: I have no commercial ties with Ayuda Heuristics