Diagnosing Digital Woes

Posted by Shammah Banerjee | 05-Apr-2019 09:53:03


The NHS’s interaction with artificial intelligence has been met with a whole variety of reactions. While commentators from the national newspapers herald an “AI revolution” in healthcare, reports from the NHS suggest that the organisation’s infrastructure is not yet ready to take on these new developments.1

Nonetheless, the excitement around AI in healthcare is wholly justified: recent medical breakthroughs have used AI to diagnose skin cancer from pictures of moles, eye disorders from retinal scans and heart disease from echocardiograms.2 In many of these cases, AI is able to make the diagnosis faster and more accurately than a human.

The implications of this are, obviously, huge. With an ageing population and no sign of increasing funds for the NHS, the healthcare system desperately needs technology that can scoop it out of financial deficit. Its current infrastructure, Health Secretary Matt Hancock commented, can result in ‘sub-optimal care [for patients]...and ultimately lives lost’.3



At the GP recently, the lady in front of me was trying to order a repeat prescription from a receptionist struggling to find her record. Defeated, she suggested booking another appointment with the doctor. ‘Another?!’ the lady replied exasperated. It’s not a case of incompetence, but rather outdated infrastructure in a digital world.

For Indra, “digital” shouldn’t be considered a novelty in health.

‘You don’t say “digital banking”, do you?’ she said. ‘You don’t say “digital travels”. “Digital” is part of our nature now and I think we need to start driving to say let’s not categorise this as “digital health”. It’s just health delivered in a digital way. We’re creating a portfolio called “Empower the Person” which is all about equipping citizens with technology. The idea is to make the NHS a little more digitally acceptable and give citizens a clear journey through it.’

In a service used by 65 million people, being “digitally acceptable” has a lot to do with data. NHS England’s former Chief Digital Officer, Juliet Bauer, articulated the need to move from ‘the traditional “one-size-fits-all” approach…towards much more personalised and targeted interventions”.4 For Indra this is where AI comes in.

‘There are lots of purists and academic debate on how you define artificial intelligence, but one might argue that AI is really a form of maths and statistics,’ she told me. ‘I think we’re far away, at the moment, from seeing artificial intelligence implemented in healthcare. What we are seeing is the ability of computers or algorithms to spot patterns that the human eye cannot see or crunch data in a way that humans find difficult.’

‘It’s about creating efficiencies in the system,’ Indra told me. ‘‘Take, for example, imaging. If you request a CT [scan] in an emergency setting, there is a lot of waiting around, often with someone in pain or discomfort. In this setting, AI or machine learning algorithms can help create efficiencies and enable better care by giving a decision in real time whether that image is normal or not.’



Excitement around AI in the medical world has led to huge success for companies like Ada and Babylon Health (whose Clinical Lead spoke at Nimbus Ninety’s recent summit Chief Disruptor LIVE). January 2016 saw a frenzy of investment in Babylon, with the company raising $25 million in a series A funding round.5 Ada similarly boomed in 2017 with $47 million raised from investors. Indra, however, seemed sceptical of their success.

‘Symptom checking is not new,’ Indra said. ‘A lot of these platforms say they symptom check – but that’s been happening for a long time. It’s really about collecting large amounts of data. The probabilities of something happening are higher because it’s got more data to work on, it’s more accurate. But that is only ever as good as the information that goes into it. If you’re telling a 24-year-old girl that her tummy pain is period pain but actually all your data was based on 24-year-old males – maybe think about it.’

In the case of medical chatbots, one of the dangers of homogenous data is misdiagnosis. At the end of 2018, a paper published by a Brown University professor revealed concerns around the diagnoses made by Babylon’s chatbots.6 While it’s hard to pinpoint the causes behind misdiagnosis, an article in Forbes claimed that 10-15% of Babylon’s chatbot diagnoses had “missed warning signs...or were just flat-out wrong”.7

With the NHS including the ‘111 Online’ symptom checking feature in their own app, there have been concerns around how this affects patient wellbeing. Both Babylon and the NHS responded to Forbes’ comments with rebuke: their technology was meant to enhance existing healthcare systems and alleviate pressure off it – not to completely replace it. All chatbot diagnoses must be mediated by common sense.

Indra agreed. “We need to effectively educate people and ensure we create intelligent customers. That involves helping them understand to consult a healthcare professional if the symptoms haven’t gone away. There are urban legends of Google Maps showing a road but in reality there’s a cliff edge. As the human, you need to look out of the window! Don’t just blindly follow a thing.’

‘We need to disrupt the healthcare model. We’re doing it slowly by saying [to patients] here are a bunch of tools; do as much as you can, until you need help from us.’




  1. NHS. (2018) Accelerating Artificial Intelligence in health and care: results from a state of the nation survey London: Department of Health and Social Care.

  2. Sample, I. (2018) ‘“It’s going to create a revolution”: how AI is transforming the NHS’. The Guardian [online]. Available: https://www.theguardian.com/technology/2018/jul/04/its-going-create-revolution-how-ai-transforming-nhs [accessed 10.02.2019]

  3. Ranger, S. (2018) ‘NHS and technology: Making the case for innovation’. ZDNet [online]. Available:   https://www.zdnet.com/article/nhs-and-technology-making-the-case-for-innovation/ [accessed 08.02.2019]

  4. Bauer, J. (2018) ‘Transforming healthcare in the digital age’. NHS England [online]. Available: https://www.england.nhs.uk/blog/transforming-healthcare-in-the-digital-age/ [accessed 18.02.2019]

  5. Morrison, C. (2016) ‘UK digital healthcare firm Babylon raises $25m in series A round from investors including Innocent Drinks and DeepMind’. City AM [online]. Available:  http://www.cityam.com/232337/uk-digital-healthcare-firm-babylon-raises-25m-in-series-a-round-from-investors-including-innocent-drinks-and-deepmind [accessed 06.02.2019]

  6. Fraser, H. and Coiera, E. and Wong, D. (2018) ‘Safety of patient-facing digital symptom checkers’. The Lancet [online]. Available: https://www.thelancet.com/journals/lancet/article/PIIS0140-6736(18)32819-8/fulltext [accessed 06.02.2019]

  7. Olson, P. (2018) ‘This health startup won big government deals - but inside, doctors flagged problems’. Forbes [online]. Available: https://www.forbes.com/sites/parmyolson/2018/12/17/this-health-startup-won-big-government-dealsbut-inside-doctors-flagged-problems/#47a1d126eabb [accessed 06.02.2019]

Written by Shammah Banerjee

Shammah is the Senior Editor at Nimbus Ninety. She tracks down the most exciting stories in business and tech, produces the content and gets to chat with the biggest innovators of the moment at Chief Disruptor LIVE.

Leave a Comment