Machine Medicine Technologies: A tale of AI in healthcare
Machine Medicine Technologies is a London-based digital healthtech start-up that has developed a mobile app that digitally detects the motor characteristics of Parkinson’s disease.
Lovingly named after the famous physicist, Baron Kelvin, who had a penchant for measuring things, the platform can be used on any device, be it smartphone or tablet, to record, store and analyse the motor function of Parkinson’s patients on video. Using machine learning, the software analyses a video clip of a patient’s motor function to detect motor dysfunction in Parkinson’s patients far more reliably than doctors can on their own. Machine Medicine Technologies, and Kelvin with it, is a prime example of the transformative influence of AI on healthcare.
A quick introduction to Kelvin can be found here:
How did Machine Medicine come about?
We saw a need! The greatest challenge was probably creating the team. We pieced the group together bit by bit, and now we are a team of four. There is Eleni, who does our marketing and communications. Then there’s Kunal, who is a full stack engineer; Amr, a data-scientist and machine-learning expert, and me, with a clinical background as a neurological doctor and a background in machine learning.
Did you feel like you could have a bigger impact as an entrepreneur than as a doctor?
Yes, I did. What I found working as an academic clinician is that it is unscalable. You’re a little cog in a massive machine. You might see, say, six patients, for each of which you do the same thing: you examine them, the patient walks out the door, and the next one is already waiting. If you work twice as hard or twice as long, you will do the same thing for twice as many patients, so it’s inherently unscalable and directly linear as an activity. The same is actually true in the scientific world nowadays. Science nowadays is a huge machine, and a typical researcher is churning out papers every 6 months. Again, if you work twice as hard or twice as long you’ll end up churning out twice as many papers. Progress is measured in terms of papers published, not discoveries made or your contribution to the scientific endeavour. While I was on the standard academic clinician track, I realised I wasn’t going to have a great impact on medicine, and neither was I by working in academia. I believe that the only way to get out of this vicious cycle nowadays is to leverage the power of entrepreneurship with a company that does something like we do.
Our system can be used on any mobile device: tablets, desktops or laptops. It’s also inherently scalable. That’s why we are excited about what we are doing. We as a company can achieve real scalability, which is hard to achieve as an academic or as a clinician. Have you ever heard the phrase, “Don’t work in your company, work on your company.”? We’re basically doing the same thing; we don’t want to work in medicine, we want to work on medicine as an entity.
Why did you focus on Parkinson’s in particular?
Parkinson’s is a great place to start because it consists mainly of the clinical assessment. Once you have the diagnosis the clinician has all the information he needs to choose the right treatment, such as a change in drug dosage or the need for deep brain stimulation. The only information that is needed and can be used for the management decision is based on the visual and clinical assessment of how bad the movement disorder is, which is why we apply pattern recognition and machine learning to this.
Could the technology be adopted for the diagnosis of other disorders?
Parkinson’s is really just our first big project. As you’re expanding your business to the general strategies that you can adopt, you have the option of either going deep or wide. As a rule of thumb, if you’re doing something that is technologically very challenging then it’s usually a good idea to go deep first and try and solve the problems that present themselves. We are a business that is facing technical challenges, so we are going deep into Parkinson’s first.
How have you approached funding?
Funding is one of the most challenging things of being a MedTech start-up. There are all sorts of barriers to making your first sales and going to market. Raising funds is a challenge, but what we’ve done is to go to people who understand the market, who have been working in the field and could guide use.
We’ve also used a grant from Innovate UK, which helps greatly with de-risking the proposition. That is government money. Investors are always scared of funding a research project that ends up having no viable commercial outcome, so programs like Innovate UK can really draw the strings for Venture Capitalists to give three to four times more money by greatly reducing their risk. There are a lot of people out there who have Parkinson’s disease, roughly 10 million. A crowdfunding platform allows us to access these people, their family members and people interested in Parkinson’s in a way that VCs could never be.
Funding is always challenging, so all you can do is to stick at it. Tenacity and perseverance are the name of the game.
How difficult is it to integrate new AI technologies into large health care bodies such as the NHS?
Very, very difficult. That’s why we are not trying to do it personally. If you do want to interact with the NHS or the trusts that make it up, you should do it through bigger companies such as Siemens or Fujitsu. One of our goals is to integrate our technology into large companies such as Abbots and Boston Scientific. That means we deal with them, and then they will handle insurers in the US and the NHS. In contrast to us, they have the opportunity to speak to the right person quickly. We tried, we went to several hospitals that might have been able to benefit from our help, but the inertia in the system is strong. If you can find a way to get around it, I would strongly recommend that you do so.
Where in the health care sector do you think AI will have the biggest impact?
AI and computational statistics are able to scale clinicians. They allow clinicians to see dozens of patients instead of one, because they won’t have the administrative component that they have at the moment of keeping handwritten or digital records. They’ll be able to get results, see trends and look at visualised data that today is just impossible with the use of [the NHS’s] outdated 1990s pile systems. In the next ten years, AI will probably be able to transform any area of medicine that relies heavily on pattern recognition.
What is your vision for Machine Medicine 5 years from now?
Our plan is to have a platform that, like a medical device, is used to automatically assess patients and provide feedback to systems such as Deep Brain Stimulation so that they can be optimised and personalised for individuals. We also expect to be working on numerous other strategies by that time. One example would be Shunt monitoring in patients with hydrocephalus. The shunt is basically a tube that allows excess brain fluid to exit the body via the abdomen. About 1% of patients with hydrocephalus has such a shunt, and almost 100% of those will have a block of the shunt at one point. Picking up that block early is crucial, so we expect to be involved in systems like that. By monitoring these people with shunts we can pick up motor abnormalities that herald an occluded shunt. Another example would be to pick up neurodevelopmental disorders in children early on. Those are the some of the adjacent markets we expect to be working on in five years’ time. Ultimately, we want to be a behaviour and analytics platform that’s used in surgical theatres as well as in people’s homes if they need monitoring.
What advice would you offer to young entrepreneurs looking to start their own biotech/AI start-up?
Be prepared to suffer! [*laughs*] It’s pretty hard, but it’s also incredibly cool. If you are willing to live on a crappy salary for a while, the rewards can be great and you are able to do something amazing. There are different life sciences support hubs depending on where you are based. But it differs depending on what you’re doing. We are London based, and I would say that it’s hard to beat London in terms of support, even though there are other support systems for start-ups in Bristol, Edinburgh etc. If you need cheap labspace, it might be a good idea to go for example to Nottingham.
The Science Entrepreneur Club (SEC) is a non-profit organisation of curious minds that aims to explore and unite the life science ecosystem by educating, inspiring and connecting. We give scientific entrepreneurs a network and a platform to showcase their innovative technologies, find investors and accelerate their company.