Home PublicationsData Innovators 5 Q’s for Ben Maruthappu, Co-Founder of Cera

5 Q’s for Ben Maruthappu, Co-Founder of Cera

by Nick Wallace

The Center for Data Innovation spoke to Ben Maruthappu, a medical doctor and co-founder of Cera, a UK-based company that uses artificial intelligence (AI) to deliver elderly care services. Maruthappu explained why choosing the right level of care for an elderly person can be a complex decision, and how AI can help.

Nick Wallace: You are a medical doctor, and you also have advised NHS England on its innovation policy. What got you involved with AI, and how did you come to establish Cera?

Ben Maruthappu: I advised a venture capital fund in the United States on their healthcare investments, where I came across numerous health tech companies. I practice as a doctor—including in geriatrics and emergency medicine—and spent three years advising the chief executive officer of NHS England on technology, and leading a lot of technology rollouts, setting up a number of our key programs that look to scale innovations across the NHS, and then I decided to co-found Cera.

The reason I founded Cera was, firstly, I had to experience it myself, organizing care for a loved one. Secondly, as a doctor, I could see the problems of bad elderly care and how that had a tremendous effect on patients as they ping-ponged in and out of hospitals because of the low-quality care they were receiving at home. If they received high quality care, they wouldn’t be coming into hospitals so much, and they wouldn’t be deteriorating so often.

Also, at the national level, I could see the knock-on effect that a dysfunctional care system was having on the NHS. Around 4,500 patients per day are stuck in hospital because their care has not been organized on time. This has huge consequences for the NHS, it’s bad for patients, because they’re stuck in hospital when they could be at home, and it causes local governments to incur financial penalties and receive fines because people are not discharged on time.

So all the way from the individual, personal level, to the national policy level, I saw the issues that a poor care system had. But then I also realized through my exposure to technology that actually a lot of these could be solved if we implemented the right form of technology and the right form of digital to transform the sector. So that’s why I decided to dive into co-founding Cera.

Wallace: What can AI do to improve elderly care services?

Maruthappu: We are a technology-enabled home care provider. We organize for carers to go to people’s homes, help get them out of bed, eat, drink, take medication, bathe, and so on—supporting older people to live more independently in the comfort in their own homes, as opposed to, say, going to a care home.

We use both digital and AI to radically improve the way in which we deliver services. The digital makes us more efficient and transparent, because we ditched all of our logistics, the way in which we gather information is more transparent, and so is the way we match patients and carers based on skills, language, availability, and so on. That’s where digital comes in: It makes us more efficient, scalable, and improves our quality.

The AI comes in in a slightly different manner. We realized that every time our carers see patients in their homes, they record lots of interesting and valuable information about how the patients are doing. For example, how are their health conditions doing—are they getting better or worse? Which medications have they taken? What have they been eating and drinking? How drowsy is the person? What’s their urine output? We run AI on this data that we collect, and we use it to predict if our patients are likely to deteriorate.

The AI is important, because say someone has early symptoms of a urine infection, or early symptoms of their blood pressure getting worse: We can identify that on day one or two of the deterioration, as opposed to it getting much worse and let’s say on day five or six of this person having an infection, them having tremendous symptoms, becoming unconscious, and having to go to hospital. If we can identify the problem earlier, through use of AI on our data, that means we can act much more quickly, it means we can increase the amount of care someone receives or contact their doctor, it means that they can receive the right services and the right treatments earlier on and be healthier, it also means we can potentially avoid people having to go to a hospital, and potentially dying there.

Wallace: What patient data matters most, and do you use any other kind of data, such as medical research or other publicly available information?

Maruthappu: At the moment, we use purely internal data, and that’s because our carers are seeing these people three or four times a day, so we have a very high resolution database. When we onboard patients to our system for the first time, we run full diagnostics where we record all of the conditions they have, any medical issues they have, and any other important pieces of information that are required. All that goes into our system, but we aren’t integrated with third party data sources. It’s completely internal. At this point in time, we don’t need to link up to other data sources.

However, we are currently in discussions with the NHS, specifically looking at primary care, and at how medical records could be linked to the care records that we are collecting. That would mean carers could access medical records, and doctors could access care records, which would arguably put them in a better position when making decisions about analyzing a patient’s health.

As for the patient data, we collect a whole host of data, and it’s the breadth and depth of the data that’s so important. We collect data on everything from just how much a patient is urinating and when they’ve opened their bowels, the medications they’re taking—or not taking, for whatever reason—to what they’re eating and drinking and how much liquid they’ve ingested, to what they’ve done in their day. If they’ve got particular symptoms or conditions we’ll note things like their blood pressure, their blood glucose, if they’ve got a swelling we’ll note how bad it is and if it’s changing over time.

This is a combination of structured data and unstructured data. We use artificial intelligence to convert unstructured data into something that we can actually count and analyze, and then we use analytics on the whole dataset—unstructured and structured combined—to start making predictions and risk assessments. We are able to converge a whole host of data metrics and look for subtle changes in certain things we’re collecting, which can indicate that somebody might be at risk or on the verge of deteriorating.

For example, a carer could record something such as, “Mrs. Smith is currently feeling a bit more drowsy than usual,” but this could be a normal variation. There carer won’t be sure: Is this just Mrs. Smith being more tired than usual, or is this the early signs of an infection? When you look at the historical data that we’ve collected about that patient, you can tell what the early symptoms are of when they’re going to deteriorate, and you can see that, actually, the last three times Mrs. Smith had to go to hospital, she became drowsy just before she did that. Those are the subtle signs and signals that an analytics platform can identify, but a human would find difficult to identify.

Wallace: How does data produced by devices connected to the Internet of Things enhance the capabilities of AI tools for use in elderly care?

Maruthappu: That’s something we have in the pipeline for this year. The potential value of this is significant—I don’t think people have really harnessed the potential value of connected devices. If you can have a wearable on an older person that measures their vital signs and movements continuously, and combine this with delivery of home care services, that could be a really valuable combination, because it means you can remotely monitor people, and if something happens you can send a carer in on demand to check if everything is okay. But apart from that, it means you can potentially reduce the amount of hands-on care someone requires in a given day, which makes it much more cost effective for them when they are purchasing home care services. It also means a carer can see six or seven people in a day as opposed to two or three. Connected devices put us much more on the front foot, they allow us to identify very small changes in vital signs and other metrics, and they allow us to reduce the amount of care and the cost of care that someone needs to receive as well.

Wallace: Where do you see AI in healthcare going over the next few years?

Maruthappu: I think AI in healthcare is not just possible, it’s inevitable. It’s something that will definitely happen. We’re starting to see it bubble-up now, but in five to ten years, I imagine that AI will be prevalent across health systems, be it supporting radiologists in analyzing images, be it supporting clinicians and doctors in diagnosing patients, or in our case, in identifying patient deteriorations earlier on and preventing them from even happening. I think multiple parts of healthcare are going to be transformed by AI in the next few years, and it will be commonplace. In the same way that the smartphone over the past decade has become widely used by most healthcare staff, over the next decade I think we’re going to see a similar movement when it comes to AI. I think indirectly or directly, most people in healthcare will be using AI.

AI is most impactful where you have large, knowledge-based sectors, because it’s very hard for a human to absorb and act on all of that knowledge. Medicine is a good example of that. There are new facts, new guidelines, and new medications coming out continuously, and it’s very hard for a human—be it a nurse, a doctor, or another healthcare professional—to keep up, whereas with an AI platform, the more it learns and the more it sees, the better it becomes.

AI can be an expert in multiple specialties at the same time, it can be an expert in cardiology and respiratory and gastroenterology at the same time, unlike humans, which need to be an expert in just one of those. Not only can it be an expert in multiple specialties, at the same time it’s scalable, so you can have thousands or tens of thousands of people using the same platform at the same time, and the more people who use it, the better it becomes. Those are two factors and strengths of AI that make it extremely appealing in the healthcare sector.

You may also like

Show Buttons
Hide Buttons