First published Winter 2018


In 2018, the NHS gave the green light for GPs to prescribe the GroHealth low-carb programme for type 2 diabetes. Louise Wates spoke to Arjun Panesar, one of the brains behind its creation


Although basing our diet on starchy carbs is still accepted official wisdom, the recent approval of a low-carb programme for patients with type 2 diabetes and pre-diabetes has demonstrated a tremendous shift in the starchy carb dogma of recent decades. Now, low-carb has officially been given the blessing of the NHS — albeit in specific medical circumstances. But it isn’t just the low-carb approach that makes this interesting news; it is the fact that using data from a large, online community has helped to inform and develop a digital programme that has the potential to help save healthcare costs around the world.

Also available to anyone without a prescription, the GroHealth low-carb programme, which comes under the umbrella of Diabetes Digital Media, can be accessed through the associated website There is a one-off charge of £29.99 for the full experience, but a wide range of resources can be accessed for free. Currently, according to the website, more than 377,000 people have signed up.

Created in part using data from, which has grown to become Europe’s largest online diabetes community, and under the guidance of GP and low-carb champion Dr David Unwin, the programme is an illustration of how artificial intelligence is paving the way towards tailor-made complementary health solutions. Yet its beginnings were humble, starting as “a bit of a student hobby” for “self-confessed data geek” Arjun Panesar.

What was really interesting for us to hear was that [empowerment] meant you had a better relationship with your healthcare professional

A family problem

It may now have members from as far afield as India, Australia, and Nepal, but was initially created only because Panesar wanted to help his grandfather.

It was whilst Panesar was studying at Imperial College, London, that his grandfather underwent heart surgery and was also diagnosed with type 2 diabetes. But having been given advice by doctors in the US, where he had initially been diagnosed, and then by doctors here in the UK, his grandfather was “kind of conflicted”, says Panesar. Although being a highly-academic man who understood atoms and energy, his grandfather was finding it difficult to work out — from the information he had been given — what diet was best for his health.

“I’d come home for Christmas and he said, ‘I still have no idea what to eat’,” says Panesar.

“He’d been keeping these logs that he’d been writing down... and he was telling me he had no idea as to whether what he was eating was good for his heart.” At the time, his grandfather was thinking about his recent heart surgery, but his diabetes was also an issue. “So I said why don’t we ask the internet?”

The domain name had previously been bought by Panesar’s father back in the ‘90s and had been left to go dormant so it was, as Panesar says, a “no brainer” to use it. But social media was still in its infancy. Although there were signs that the internet could connect people with similar interests, talking about deeply personal information such as health was mostly uncharted territory. “I remember speaking to one of my lecturers and saying do you think people will talk about their condition on the internet?” The lecturer’s response was ambivalent; yet once was launched, the forum received its first question within minutes. “It was really encouraging,” says Panesar. “It was like ‘oh wow people are actually talking’!”

Data capture

Panesar left university with a first-class honours degree (MEng) in Computing and Artificial Intelligence, and the website remained in the background “still something of a hobby” until 2010. But it had a “decent” number of users; and looking at the data, it was possible to identify what health issues and concerns commonly affected the forum’s users.

“In 2010 we were just looking at the data and it was really obvious that people’s main concerns were about food,” he says.

“We could see from the conversations that we were having that people didn’t know what to eat, so we put this cookbook together.”

The cookbook, which he describes as a “basic, 30-day cookbook” was put out on New Year’s Day 2012 and made available online. Within three days it had been downloaded 30,000 times. Within a month, 90,000 times.

What the data also showed was that it was predominantly people with type 2 diabetes who were using the site. So in 2012, Panesar and his colleagues “did a little bit more digging” and decided to carry out some research.

The team came up with a “74-question survey” with a target of 10,000 responses — undeniably a large number, particularly for a relatively young organisation. Yet they offered none of the traditional bait such as prize draws to get people to engage. The only incentive offered for anyone to sign in and sign up was the team’s promise to feed back the results to the respondents.

“What was fascinating about it in particular was there was no incentive. There was absolutely nothing for the user to gain — other than that we said we’d present the data back to the users once we reached our target of 10,000.”

Within the space of four weeks, the survey had received 20,000 responses, doubling the target figure.

“That data we had — from almost baseline — began the direction of where we are today,” says Panesar. “Twenty-twelve was when we really started looking at the data from an AI perspective... from a neural network perspective.”

Using this kind of data capture has enabled the team to see the types of concerns that people have when they first join the forum and compare it with the concerns that they have six months or a year later. Since 2012, the team has also collaborated with academic institutions to find out whether what it does has a measurable impact on health outcomes.

Patient-GP relationships

Panesar says that in 2015, research conducted by Royal Holloway College, University of London, concluded that using the forum had an empowering effect for individuals when they then spoke to their GP. “What was really interesting for us to hear was that [empowerment] meant you had a better relationship with your healthcare professional.

“That was interesting to us because five years ago we absolutely did not see that. Five years ago we saw that around 80 per cent had a worse relationship with their healthcare professional as a result of being part of the digital community. Today we see it completely the other way round — so we’ve kind of seen a pendulum swing.”

Rather than seeing this improvement in patient-GP relationship as stemming from a change in attitude within the medical profession, Panesar believes it is because patients are better informed. He admits that the patient experience can depend upon what clinical commissioning group people fall under, but the data also reveal that some people are now being referred to the forum by their GP.

The aim of the forum, however, he stresses, is not to replace the care given by GPs. “AI and digital technology is not here to replace humans. It’s here to augment humans — so one of the things that we are really keen on doing is facilitating that relationship between patients and their healthcare provider. We don’t want to replace the relationship, we just want to make it a little bit easier.”

Blended healthcare

But using AI for what he calls “blended healthcare” is also still in its infancy and so hasn’t yet revealed its full potential. “Because a lot of what we do is digital and because a lot of what we do is... novel is a better word than innovative... these are new paradigms of care that typically haven’t existed before,” he says.

“We’ve taken lessons from the forum. We know that after six months of being on the forum people’s blood glucose levels improve, and we know that their diet choices improve just as a result of being part of the forum. So we’ve almost taken that evidence from the forum and put it into the low-carb programme in order to create future concepts of health.” the future of AI is still unfolding, it is this that Panesar hopes will be the paradigm for the future – complementary, blended healthcare working with the NHS

Affordability and sustainability

Whilst recognising that a low-carb approach is still somewhat political, Panesar says that it has been demonstrated that it does produce results, is affordable and sustainable. “We’ve kind of entered the political game as an organisation by the sheer nature of calling it the low-carb programme... it’s not about being in ketosis because it’s about blood glucose normalisation... in reality, it could be called the ‘eat real food avoid processed crap programme’.

“We’re not a no-carb programme, we’re a low-carb programme. We’re not telling people to eat particular foods, but what we do know from the data is that people are eating four or five times the recommended daily allowance of carbohydrates — and as a result are taking in so much energy that they’re obviously not expending.

“Regardless of whether you use a ‘calories-in calories-out’ approach or low-carbohydrate approach, people are still taking in too many calories.”

As a tool to help people better manage their health, the programme is not being presented as a miracle cure. Success depends upon the individual user — and on the forum it is certainly possible to find conversations about “burn-out”, which is when people get tired of trying to manage their condition and need extra support.

Panesar says success can also depend upon how well the user can relate to the programme. Working with Punjabi GP Dr Kesar Sadhra, the forum is already better able to connect with south Asian users.

Like Drs Sadhra and Unwin (see Optimum Nutrition Autumn 2018), both of whom adopted low-carb diets when working with their own patients, Panesar also follows a low-carb diet. “You think you’re not eating badly but when you begin to count the number of sandwiches and the unnecessary insulin being released in all of our bodies…” He adds that everyone who works for the company has changed their diet to some degree. But, coming full circle, it has also dramatically impacted his family’s health.

“My Indian grandmother was pre-diabetic for about eight years and about two years ago moved over to the type 2 phase. She was so concerned, having seen what we’ve been working on, she essentially took it as a terminal diagnosis of a disease and was probably in grief, I would say. My dad had what was probably a typical Indian attitude of ‘life is going to happen, so let her enjoy her time’. But what I was saying to my parents was: ‘Can’t you see what we’re doing with all these people? All she needs is support?’.

“What was probably one of the biggest motivators for her was one of her friends had a stroke that was related to type 2 diabetes and got progressively worse — and my grandmother watched that... then she spoke to me and said: ‘What do I need to eat?’. We went through a list and she’d say, ‘but I really like this’... So I’d say all right then, how about you just don’t eat it until 12 o’clock or just eat it until one o’clock.”

The concept of intermittent fasting works well with Asian culture, Panesar explains, but also demonstrates the importance of any programme being culturally relevant. And as the future of AI is still unfolding, it is this that Panesar hopes will be the paradigm for the future — tailored, complementary, blended healthcare working with the NHS.

But being a trailblazer does have its problems. “The difficulty is that you have very few peers in this, and the tricky thing is to make sure you’re doing it right,” he says.

Yet given the success of the programme so far, Panesar and his team must certainly be doing something well. Not bad for something that started as a bit of a student project.



Read more articles and recipes