“The main concern of using AI in healthcare is ethics: who owns the data and what is it used for?” - Arjun Panesar, CEO, Diabetes.co.uk

Over 500 attendees will come to AI Tech World on the 29th of November at Olympia Conference Centre to see over 60 leading speakers discuss the impact of AI on Business.

One speaker, Arjun Panesar is CEO of Diabetes.co.uk. Arjun has a decade of experience with intelligent health systems and big data. With a Masters in Artificial Intelligence from Imperial College London, Arjun's focus is transforming healthcare through empowering patients. The Low Carb Program demonstrates significant health improvements - including the reversal of type 2 diabetes, a condition once understood as chronic and progressive. Through the use of innovative digital engagement, real-world data and genomics, Diabetes.co.uk is reshaping the understanding of wellness.

Arjun, who will be discussing ‘Machine Learning in Healthcare’, completed the AI Tech World speaker Q&A:

 

1.       How are you using, or planning to use AI in your business?

I lead the team at Diabetes.co.uk – we are the world’s largest and most engaged diabetes community – with over 600,000 members. Having engaged with this community over the last decade, we have developed a number of digital health interventions, or digital therapies, that incorporate a considerable amount of AI.

Our Low Carb Program (for people with type 2 diabetes, prediabetes and obesity) collects over 200 variables per patient and gathers thousands of data points daily – both structured and unstructured. We’ve created a number of intelligent algorithms that are able to learn what success on the Low Carb Program looks like for each particular user. What this means is, we can learn from the successes of other people in similar positions, to present an optimal engagement pathway which empowers patients to achieve their health goal. What’s even better is, we can do this in real-time!

A common question is how the success of AI applications like this can be measured. A good measure of success is impact - at a year, 70% of people remained engaged with the Low Carb Program with 35% of people reducing the number of medications they were on.

 

2.       What benefits do you expect to see from using AI?

The ability to make robust, instant decisions with (almost) 100% confidence. Any piece of AI trained on a substantial amount of data typically enables us to make generalisations. And, as AI intelligence and data validity improves, you’d expect these decisions to become even more precise. These generalisations that can unlock a wealth of opportunity – as they can be used in other situations where data is missing. In our own Low Carb Program, we were able to learn that if people were unemployed and under or over a certain age, they were more likely to suffer from diabetes-related depression or distress. What was even more useful was that we could then predict the likelihood of this (with a 96% success rate) and present a different experience to that member.

 

3.       Do you have any concerns about implementing AI?

The main concern of using AI in healthcare is ethics: who owns the data and what is it used for? Everyone’s used to posting on Facebook or Twitter but when it comes to sensitive data, people naturally become cautious. When it comes to AI, most people are concerned that we could learn something that could have a negative influence on something else.

We found that approaching AI and what we do with our data in an ‘openness environment’ had a significant effect in people’s understandings. We let users know what their data can be used for, how it helps improve their experience and how it can be used on a wider scale to affect global health. We have found that approaching data transparently is by far the best policy.

Taking that further, what if we could learn something that you didn’t want to know? We have developed a system that can predict the likelihood of pancreatic cancer and the ethical challenges of disclosing that information are significant.

Finally, one of the challenges around AI and in particular machine learning is that algorithms are picking up deeply ingrained race and gender prejudices concealed within the patterns of language use. Overcoming bias is key to being able to use AI to its true potential. 

 

4.       What do you consider to be the major challenges?

The biggest challenge around AI is the sheer variety of data available – and being able to process it efficiently. People are generating tremendous amounts of data – both structured and unstructured, through their phones, smartwatches, apps and IoT.

In the medical industry, one of the fundamental challenges is addressing the variety of data at hand and being able to use that data. Hospitals generate huge amounts of patient data, so do doctor surgeries; but it’s disconnected from specialist care, patients’ own data and we can’t even access our health records digitally. No one organisation, or even the person that owns the data, has access to all of it.

Bringing all this data together would be of huge benefit – as we’ve only really scratched the surface of AI and deep learning in health.

 

5.       How do you think trust around AI can be generated?

Trust will come through demonstrating the real-life benefits that AI can bring. HSBC are testing a new app that nudges you if you overspend. That’s a great idea – with positive outcomes. DeepMind did some great work with the NHS. Clever uses of AI like this help to demonstrate the positive impact data and AI can have. People were interested to know why we wanted to collect insulin medication details in the Low Carb Program. When we explained how people with type 2 diabetes were reducing their medication dependency and how others on insulin would receive a different experience, people became more open to sharing their data. It’s about being clear, honest and transparent with what will happen to their data and to what extent it will help them. After all, it’s their data!

 

6.       What impact could AI have on your customers’ experience?

It’s single-handedly changing the lives of our members. Each patient that completes the Low Carb Program saves the NHS more than £1,000 per year – just in medication deprescription costs. Type 2 diabetes - a disease that was once considered to be chronic and progressive - is being reversed in thousands of people. It’s saving the NHS millions of pounds, rewriting the medical textbooks and redefining the health landscape. The impact of AI in real-life, for our members, is truly life-changing. It’s enabling people to live longer, happier, and ultimately, healthier.

 

7.       Which industries could see the biggest benefit from AI?

Health. The finance, energy and transport industries have been quick to integrate significant amounts of AI. However, the health industry has only recently seen a surge – perhaps only really over the last 5 years. This makes sense - lives are at risk in healthcare, so you can understand why innovation has been slow to propagate. AI is only as good as the data it’s fed, so ensuring data validity and consistency in healthcare is paramount.

Pretty much every industry you can think of is integrating some form of AI. Fridges can self-replenish stock and cars can drive themselves. With so much data being generated that ever before, and with phones now having more power than Apollo-era NASA computers, there is a visible, exponential growth in the application and use of AI.

8.       To what extent do you think machines could replace humans in the workplace?

It depends in what capacity. Robots have been replacing jobs for years - a lot of packing and logistics is taken care of through robotics.

This year in Japan, Fukoku Mutual Life Insurance made over thirty employees redundant and replaced them with IBM’s Watson Explorer AI. Humans are prone to make mistakes – so for tasks that can be computerised sufficiently, replacing humans with robots is a very real possibility.

In the medical setting, it’s unlikely life and death decisions will be handed to a robot anytime soon. Humans are multi-dimensional when it comes to health and determining individual treatment pathways. Although there are significant advances taking place – including robotic operations on patients, when it comes to making patient diagnoses and identifying treatment pathways, it’s likely a human element will always be required. Why? Because computers can make mistakes too – typically influenced by human error or bias.

However, AI is only as good as the data that it learns from and the application setting. Will robots ever rule the world? Probably not. Could they? Well, it may depend on the autonomy they are given. The ethics of autonomous agents and AI is very interesting, can a robot ever truly ‘feel’?

 

Can AI replace humans? Certainly, they already are - but perhaps not in every way. 

 

About AI Tech World

AI Tech World’s unique agenda has been specifically created to maximise the busy schedule of the senior executive. 

In only one day out of the office, hear from visionary keynotes, relevant case studies, and get to grips with the subjects that matter most to your business with our “how-to” industry-specific Labs. With over 60 leading speakers, come and discover exactly how industry visionaries are breaking new ground within Artificial Intelligence.

Amid unrivalled content, AI Tech World London offers exceptional networking opportunities for attendees to connect with industry leaders and innovators. Do you have a burning question or need something clarified? Use our “Ask the Experts” service to ensure you come away with all the knowledge you need to help redefine your company’s technology boundaries. 

Click here to register now to reserve your place for THE AI event of 2017.

For more information please call Georgia Deery on +44 (0) 330 335 3900 or email This email address is being protected from spambots. You need JavaScript enabled to view it..