The NHS, a hot bed for AI health innovation

Much is talked about health care distribution channels such as payers here in the US. Yet just across the atlantic, there is a payer that is just as important especially when it comes to testing new AI ideas – the UK National Health Service (NHS).

According to the Guardian, the NHS was born in 1948. On that day, doctors, nurses, pharmacists, opticians, dentists and hospitals came together for the first time as one giant UK-wide organization.

That UK wide organization has health data records for an estimated 66 million people and it’s all standardised data – not fragmented, no interoperability issues and it’s longitudinal – going back decades. As a single payer, every one’s record is linked to a unique identifier across practices, hospitals and providers for each visit, drug prescribed, procedures and clinical notes.

This makes the data so useful to large technology companies for AI inventions especially those that are focused on predicting disease.

Here is a list of tech companies working with the NHS.

Alphabet’s DeepMind is the first to develop a product based on NHS data. In 2016 it launched a phone app, Streams, that now have been proven to detect early signs of kidney failure in patients. The app continually pulls together medical information and sends alerts to doctors if the results suggest a patient could develop acute kidney injury.

Ama­zon’s voice as­sistant Alexa will pro­vide users in the UK with med­ical in­for­ma­tion about their symptoms. Un­der the NHS part­ner­ship, which started in June this year, Alexa will au­to­mat­i­cally search medical information web­sites for rel­e­vant in­for­ma­tion when probed about com­mon ill­nesses.

Babylon Health, a UK company was recently in the news for raising $550M, on a $2B valuation, one of the largest single raise ever for a digital health company.

That raise happened off progress made on NHS data. The company originally built an AI chatbot that checks symptoms of NHS patients and then refers serious issues to a doctor. To meet the demand of the chatbot, it found a workaround creating it’s own doctors practice called GP at Hand, paid for by the NHS, which offers telemedicine among other services.

There are other US technology companies exploring work with the NHS even with the challenges of the politics around privatization and issues of patient consent.

The organization which spends approximately $150B a year is so valuable to the US that it may be on the negotiable table for a US-UK trade deal after Brexit.

Digital Health IPOs in 2019

Livongo and Health Catalyst made the public markets late last month, showcasing how the market in terms of VC and customer acquisition for B2B healthcare and the wider tech industry will likely adapt.

On the surface both companies may look like healthcare companies but on further observation, they are first, technology companies. This Wall Street Journal article shows how these companies convinced investors of the opportunity in healthcare and the opportunity costs involved in a very large market. While the sales cycle is very long, and there are regulatory burdens and requirements to meet such as when dealing with patients data, the opportunity clearly outweigh the costs. 

It’s surprising that the two companies are the first startups in healthcare tech since 2016 that entered the public markets (Side note: Change Healthcare starting trading in June although a spinout of public company Mckesson). Not particularly surprising for Health Catalyst since they deal with data and AI, a scalable SaaS proposition. Livongo on the other hand, was a dark horse. It’s not so obvious from a layman (especially classic VC types) what the company does. Wall Street journal calls it ‘hardware and software for chronic disease management’. Although that description is somewhat correct, hardware is certainly not a main feature in their pitch deck. The hardware part ie wearables is not so appealing to investors because that involves inventory, manufacturing, in many cases FDA approval and millions of dollars to build such operations that doesn’t contribute to immediate top line growth. 

On looking at Livongo’s board (flanked by insurer and major customer such as Humana), it becomes clear that the public markets is where they will thrive further and raise well needed growth capital. Yet, there were several other companies that could have gotten to the public markets much faster. In this health tech arms race, there are companies in mature verticals such as telemedicine that have raised hundreds of millions of dollars and are still not public. American Well and MD Live are great examples of this trend. Although Teladoc, their competitor went public in 2015. Even in the most mature vertical in digital health (electronic health records software market), market leader Epic is still not public, although competitor Cerner went public in 1986 and athenahealth in 2007 (then acquired and re-privatized in 2018) .

World Medical Innovation Forum: Insights on AI from healthcare leaders

WMIF Image
Ziad Obermeyer MD, Assistant Professor, Harvard Medical School during First Look session

The World Medical Innovation Forum in Boston just wrapped after three days of insights on artificial intelligence in health care. While it didn’t disappoint in respect to buzzwords as is expected in AI conferences, it was the quality of the participants and their views on AI in health care workflows that made it notable. Here are twelve quotes directly from healthcare leaders.

On AI as a capability and adding immediate value

We realized having great medical devices is not enough, we need to connect them and make them interoperable. We operate on Adaptive Intelligence rather than Artificial Intelligence. Frans Van Houten. CEO, Philips

You will need to have capability to process structured (EHR) and unstructured (genomics) data, we can say we are more advanced in AI due to our work in oncology  – aiding in virtual recruiting in clinical trials. Amy Abernethy, MD, PhD. CMO/CSO, Flatiron Health

Knowing about the range of products being developed now, in 5 to 10 years time, there will not be a single decision made in healthcare and medicine not influenced by AI. John Kelly, PhD. SVP Cognitive Solutions and Research, IBM

30-45% of each provider’s day is wasted in non-care related activities, AI can improve healthcare and reduce burnout. Peter Durlach. SVP Health Care Division, Nuance

On data generation and interoperability

We need to remove any remaining friction between insurers and providers when it comes to sharing data, we need free flow of information. Patrick Conway, MD. CEO, BCBSNC

Some doctors don’t want to be paid electronically. They are still sending paper faxes.  Roy Beveridge, MD. CMO, Humana

Pharmaceutical companies are sitting on mountains of data on clinical trials. This data is still underutilized. Jackie Hunter, PhD. CEO, Benevolent AI

On data security, privacy and consent

When thinking of AI innovation, two considerations  – medical requirements on devices as well as consumer privacy and security. Jigar Kadakia. CISO, Partners HealthCare

Facebook says we got consent from you. But then you don’t really know you signed consent. Consent needs to be clear  – in healthcare. Noga Leviner. CEO, Picnic Health

On quality and trust of all these AI algorithms 

How are we going to address quality. How is that algorithm performing and how does that impact health outcomes? Seth Hain. Director, Analytics and Machine Learning, Epic

There’s a difference between Big data and AI. Big data has informed and enabled’. What has AI done in drug development? Dr Thomas Lynch. EVP, R&D Bristol Myers Squibb

Garbage in, Garbage out. We have to de-toxify the data. Carl Kraenzel. CISO, IBM Watson Health

 

Facebook and the healthcare sector

Considering the launch of a Bayer campaign back in 2016 reported by Fierce Pharma, Facebook has been building teams that sell pharma ads to the health care sector. While still in the early stages, it is no secret that Google built those teams years ago to sell account management and ad targeting for big pharma and life sciences firms.

Google got clever however. It long invested in Google (Cloud), one of the fastest growing business segment in enterprise. With Amazon (Web Services) and Apple (Electronic health records rollout on its devices), Facebook is far behind the GAFA set in the health care industry. The recent Cambridge Analytics data privacy breach is a set back for the company in rollout of any new enterprise platforms (Workplace by Facebook) and patient data analytics experimentation, especially considering HIPAA privacy regulations. Workplace by Facebook reportedly has signed up over 10,000 organizations in a very short time and is possibly the medium to enter the industry. By working with healthcare providers on collaboration and predictive analytics tools, it may be the route to compete in the already crowded telemedicine market.

Time will only tell if the company can build products that expand beyond other people’s data and the advertising model.