In the third of a four-part series on healthcare, we focus on technology. New tech innovations promise to deliver better treatments and cost efficiencies – but concerns over data protection and flawed AI persist.
13 minute read
In 2016, DeepMind, an Alphabet-owned technology company best known for its success in creating game-playing robots, launched a collaboration with Moorfields Eye Hospital, a specialist NHS diagnosis and care unit in London. DeepMind began training its cutting-edge artificial intelligence algorithms to recognise early signs of eye diseases such as glaucoma on retinal scans.1
Two years later, DeepMind announced the first results, which showed the AI was matching – and in some cases outperforming – human doctors for diagnostic accuracy. DeepMind and Moorfields said the technology could be used to ensure patients who need urgent treatment are prioritised and treated more quickly.
But DeepMind’s collaborations with the NHS have also brought headlines for the wrong reasons. In 2017, the Information Commission, the UK’s data regulator, said the NHS Trust had broken data protection rules when passing to DeepMind 1.7 million patient records. Google has since incorporated DeepMind’s health-focused unit – previously run separately – and dissolved an independent panel that scrutinised the company’s work in healthcare, raising concerns the US tech giant will start commercialising patient data without consent.2
Google’s interest in healthcare is part of a growing trend. Other tech companies in the US and China have expressed lofty ambitions to transform diagnosis and democratise patient care. They are set to transform the landscape in healthcare for drug-makers, hospitals, pharmacies and insurance companies, as well as for doctors and patients – but questions remain about the pros and cons of Big Tech’s involvement.
Spending on healthcare is increasing as populations grow and life expectancy lengthens: total expenditure is expected to hit US$8.7 trillion in 2020, up from US$7.1 trillion five years ago.3 Properly implemented, better technology promises to offer governments ways to improve efficiency as costs escalate, and the larger tech firms see opportunities to intervene across the industry.
“The healthcare system is fragmented, with little transparency and inefficient use of resources. There is a big incumbent profit pool and customers are dissatisfied with existing provision infrastructure. So it’s understandable that Big Tech sees healthcare as ripe for disruption,” says Mikhail Zverev, head of global equities at Aviva Investors.
Big Tech sees healthcare as ripe for disruption
“In other areas this set of circumstances has created an opportunity for an innovative, technology driven new entrant: think Netflix versus the pay-TV ecosystem, Uber in personal transport or Amazon in retail,” Zverev adds.
Medical practitioners also see the potential for new technologies to improve aspects of healthcare. Eric Topol, cardiologist and author of Deep Medicine, a study of medical AI, argues the flaws in existing systems are forcing doctors to embrace new solutions offered by the Silicon Valley titans.
“The technology in medicine today is pathetic,” Topol says. “In the average hospital room, an alarm goes off 135 times a day, and 99 per cent of those are false alarms. The fact we haven’t been able to fix that over the years tells you something. So we really do need enhancements, we need every type of innovative and ingenious technology, as long as it is relatively inexpensive. We need help.”
Topol picks out Google, Apple, Microsoft and Amazon as the furthest advanced in healthcare projects, with each company focusing on a different area that plays to its individual strengths. In his view, they will have an “unstoppable impact” on the industry.
Google’s diagnostic tools build on its world-leading expertise in machine learning and AI. As well as breakthroughs in retina scanning, DeepMind’s collaboration with the NHS has also yielded an app, Streams, that alerts physicians at the Royal Free Hospital in London if a patient suffering from kidney disease needs urgent attention. According to the company, the app has reduced the average cost of admission for patients with acute kidney injury (AKI) by 17 per cent – potentially a huge saving in the context of the £1 billion annual cost of treating the condition across the NHS.4
Google also runs a separate life sciences unit, Verily, that is working with pharmaceutical companies on drug discovery and even creating devices such as ‘smart nappies’, which alert parents when their child’s nappy needs changing.5
If Google is to make any headway in consumer gadgetry it will face intense competition from Apple, which wants to personalise healthcare by giving users of smartphones and wearable devices the ability to monitor their heart rate and exercise levels. As digital sensors become more sophisticated, the range of applications will increase: the company is trialling a device that can detect the tremors afflicting sufferers of Parkinson’s disease, enabling them to monitor the progress of their condition.6
Microsoft, meanwhile, is concentrating on IT solutions, where the ubiquity of the Windows operating system among hospitals gives it an advantage; it is rolling out new software and data-storage solutions powered by its cutting-edge cloud computing technology.
As most of the tangible infrastructure is already in place – hospitals, ambulance networks, pharmacy chains – big tech companies can focus on asset-light, high-value-add aspects of healthcare, which fits neatly with their business models.
Demand among investors is strong. Research from the Harvard Business Review indicates the healthcare technology sector has a revenue multiple of 5.1x, compared with 1.3x for consumer durables, which may be one reason for the soaring valuations of tech companies making inroads in healthcare – and for the jitters among the incumbents.7
“It’s a huge growth market for the big tech companies, when you look at demographic trends; investors would do well to assess which will be the winners and losers in healthcare as more of these disruptors make their move,” says Stephanie Niven, global equities fund manager at Aviva Investors.
When you look at demographic trends, investors would do well to assess which will be the winners and losers in healthcare
Google’s parent company Alphabet reportedly has ambitions to shake up healthcare insurance in the US: by crunching patient data at a population level to highlight potential efficiencies, it believes it can undercut insurers on cost. CityBlock Health, a unit spun out of Google company Sidewalk Labs, is already providing personalised care in New York to low-income communities eligible for the government-funded programmes Medicaid and Medicare. In addition, Alphabet recently invested US$375 million to up its stake in health insurance start-up Oscar Health, which is also looking to muscle in on big insurers’ share of Medicare payments.8
Amazon is also intent on disrupting insurance, along with other healthcare sectors. In 2018, Jeff Bezos’ e-commerce giant announced a partnership with Berkshire Hathaway and JPMorgan aimed at lowering healthcare costs for the three companies’ employees. The same year, Amazon moved into the drug distribution business, paying US$1 billion to acquire online pharmacy PillPack, a tie-up that poses a threat to pharmacy benefit managers (PBMs), the ‘middle-men’ that provide links between drugs companies and pharmacies. US pharmacy chains and drug wholesalers lost US$14 billion in market value on the day of the PillPack announcement, according to Thomson Reuters data.
Circling the wagons
Companies have begun consolidating to bolster their defences against the disruptive incursions of Big Tech. In December 2018, US-based insurer Cigna announced a blockbuster US$67 billion acquisition of PBM Express Scripts, a company whose share price had been among the hardest-hit by Amazon’s entry into healthcare.9 Beyond the other operational efficiencies, this sort of consolidation should bring together useful data sets.
Healthcare companies recognise their access to data gives them a potential advantage over technology firms that are new to the industry. Take Managed Care Organisations (MCOs) such as UnitedHealth Group, which covers what it calls “the entire continuum of care”, giving it data on everything from the work of primary physicians to prescription medication to acute care. The company’s subsidiary, Optum, has created an industry-leading data analytics infrastructure that it uses to improve outcomes and create new technological tools.
“The long-held promise of data analytics and machine learning is finally coming good – the combination of depth and quality of data, performance of hardware and maturity of software tools means that valuable, actionable insights can now be generated,” says Zverev.
Valuable, actionable insights can now be generated
“We're seeing various industries adopting these tools for various purposes, from predictive maintenance of industrial assets to improved lending decisions in finance. Healthcare naturally belongs to this list, and MCOs should be among the potential beneficiaries; technology in healthcare is not only an opportunity for new entrants with tech DNA, but also for best-in-class incumbents to win through better adoption of these new tech tools,” he adds.
Flight of the BATs
Technology is also revolutionising healthcare beyond the West, most notably in China. The opportunity is sizable, as spending on healthcare in China still lags behind advanced economies. In 2016, healthcare expenditure was 6.2 per cent of GDP, compared with 10.8 per cent in Japan and 17.9 per cent in the US. Per capita spending was US$506, compared with over US$10,000 in the US.10
Hangzhou-based e-commerce behemoth Alibaba has launched a subsidiary, Alibaba Health, which is targeting pharmaceutical supply chains as an entry point into the industry. Over the longer term, it envisages a closed-loop ecosystem in which customers use its online platform to order other healthcare services.
Like Google and Amazon, Alibaba has designs on disrupting healthcare insurance. Its subsidiary, Ant Financial, has launched a platform, Xiang Hu Bao (‘mutual protection’), which covers the costs of serious conditions such as cancer. No premiums are charged, but members contribute if one of their number needs treatment for a specified disease. In the event of a claim dispute, members use a Blockchain-powered online platform to vote on whether the collective should pay up.
Insurance giant Ping An is fighting back with its own AI-powered healthcare platform, Good Doctor, aiming to leverage data and automation to lower costs. The company is trialling a telephone booth-sized unstaffed clinic in which patients can consult an AI doctor and use a smart medicine cabinet to choose a medication that fits their diagnosis.11
“Alibaba Health is about disrupting the prescription and non-prescription drug retailing model; it is taking business from pharmacies, both in and out of hospital,” says Will Ballard, head of emerging market small cap equities at Aviva Investors. “Ping An Good Doctor is more about diagnosis and lowering the cost of insurance; the first step before you can claim on its insurance for a doctor’s visit is to go through the online platform for a preliminary diagnosis and then move on to a hospital appointment. That makes it significantly cheaper than walking straight into a private clinic.”
Alibaba Health is about disrupting the prescription and non-prescription drug retailing model
Not to be outdone, Alibaba’s rival, Tencent, has teamed up with Aviva and Hillhouse Capital to create a Hong Kong-based digital insurer specialising in life insurance and critical illness cover. In mainland China, meanwhile, the company is using its near-ubiquitous WeChat platform to offer health services such as real-time text chats with medical professionals and appointment-booking services. Tencent is also rivalling DeepMind in AI-powered diagnostic imaging. Baidu – the third of the so-called BATs, China’s trio of tech giants – is developing a chatbot that draws on the formidable AI capabilities of its unit Baidu Brain.
China’s tech giants have the backing of the government, which is keen to take advantage of their mastery of data analytics and AI to improve the healthcare system and expand access among its vast population. There are, for example, only 20 eye doctors for every million people in China, making remote, tech-powered diagnosis a priority.12
Relatively loose regulations surrounding data protection allow Chinese companies access to masses of patient data that they use to train algorithms and offer technological solutions to healthcare systems management. For example, Guangzhou Second Provincial Hospital in southern China is using 300 million patient records to train AI, identify patients using facial recognition software and suggest diagnoses via WeChat.13
Access to data may give China’s tech giants the edge over US firms, where controversy surrounds their attempts to use patient information. Facebook drew negative headlines in 2018 when it reportedly requested millions of anonymised patient records from hospitals, which it hoped to match to its own user data.14
Questions over the ownership of medical information, and the balance between data privacy and medical efficacy, are nothing new. In 1951, an African-American woman named Henrietta Lacks underwent a biopsy procedure at Johns Hopkins Hospital in Baltimore, where her cells were discovered to be ‘immortal’, meaning they can reproduce indefinitely. Long after her death, Lacks’ cancer cells are still being used for crucial medical research – some of it for commercial purposes – even though she gave no consent and received no compensation.
The debates over the commercialisation of personal information have been given a new edge now heart rates are digitised on smartphones in real time and individuals are sharing their DNA with companies that may pass it on to third parties – whether they are insurance companies or law enforcement agencies. This means winning consumer trust will be particularly important for tech companies with an interest in healthcare.
According to a recent survey from Rock Health, an incubator that invests in digital health start-ups, only 11 per cent of people said they would be willing to share their health data with a tech company, compared with 72 per cent for doctors and 49 per cent for health insurers. Of those, Google was the most trusted, with Facebook lagging behind.15
Which tech company would you share your data with?
Bias and scandal
There is also the problem of technological bias. Deep learning – the basis for much of the AI technology in healthcare – works in much the same way humans learn, by reviewing existing bodies of knowledge and spotting patterns. This means human bias can infect algorithms in the same way that it can influence the behaviour of people.
Women have been traditionally underrepresented in the kind of medical trials that provide the source data for AI
Women have been traditionally underrepresented in the kind of medical trials that provide the source data for AI, which can skew the results. Take DeepMind’s much-trumpeted success in predicting forms of kidney failure among patients: look deeper, and you see the data used to train the algorithms was 93.6 per cent male, and the company conceded its model proved less effective in anticipating problems among female patients.16
As well as gender bias, AI can replicate racial bias: the images being used to train machines to spot skin cancers feature predominantly lighter skin tones, hampering its applicability in real-world situations. Other healthcare algorithms can seem comically limited: AI recently developed to recognise signs of dementia using the human voice generated optimism until it was found to have a fatal flaw – it was confounded by anyone who wasn’t an English-speaking Canadian.17
Even if the issues surrounding data ownership, privacy and bias are overcome, high-profile failures in consumer-focused healthcare tech are reason for caution. In November 2018, Google’s Verily announced it had discontinued work on a contact lens designed to test glucose levels among people with diabetes. Despite media excitement surrounding the project, the technology was found to be unworkable due to difficulties in reading sugar levels through tears.
Then there is the infamous case of Theranos, the San Francisco-based start-up that announced it had invented a miniaturised device that promised to revolutionise blood tests. At its peak in 2013, the company was worth US$10 billion. But after an investigative journalist found the company was exaggerating the efficacy of the technology, Theranos collapsed. In 2018, the company and its founder Elizabeth Holmes were charged with fraud by the Securities and Exchange Commission.18
Eric Topol says regulation surrounding the activity of tech companies in healthcare is needed to allay concerns surrounding data privacy and security; he points to Europe’s General Data Protection Regulation (GDPR) as a step in the right direction. But he is adamant that, used in the right way, cutting-edge technologies such as AI and the Internet of Things could improve healthcare.
One example, already being trialled by the NHS, is the use of voice recognition to enable doctors to spend more time interacting with and listening to patients, rather than spending an appointment inputting data and calling up records on their computer monitors. (Doubts remain over the inclusivity of this technology, given that voice recognition tends to perform worse for women and non-white people.19)
Free doctors up to prioritise urgent cases and offer more humane, patient-centred treatment
Topol also points to the potential for AI-driven diagnostic scans; not to replace doctors but to free them up to prioritise urgent cases and offer more humane, patient-centred treatment – what he calls ‘deep medicine’.
“Deep medicine is about giving time back,” he says. “The gift of time is derived from the efficiency and productivity of AI. It’s about leaning on machines to assimilate a person’s data, because that takes a lot of time and data is often dispersed between many sources. It’s also about putting patients in charge of their own data and giving them algorithms to interpret it, especially when it comes to routine, non-serious illnesses: Treatment of those could become autonomous.
All of the things that should be the essence of medicine have been largely lost over recent decades
“There are many ways to ensure that when doctors and patients come together, there is the chance for real communication, trust, compassion, empathy. All of the things that should be the essence of medicine have been largely lost over recent decades,” Topol adds.
As big tech companies continue to make progress in healthcare technology, two visions of the future emerge. In the first, individuals can track their own health using voice-activated apps, and seamlessly diagnose, as well as order online treatments for, routine ailments. When they need an in-person appointment, the doctor will be freed up by the application of powerful AI technology to spend time with the patient and offer thoughtful, humane care.
The alternative future is that companies rampantly commercialise patient data without permission, ruthlessly targeting advertisements at ill patients and tailoring health insurance products that hike premiums for at-risk groups based on their genomic profiles. AI is used not to make doctors’ lives easier, but to speed up their work so that they can pack yet more appointments into each day; patients remain entries on a spreadsheet, rather than being treated as real people.
Regulators, companies and individuals all have a role to play in determining which of these scenarios comes to pass. But one thing is certain: the rise of technology in healthcare is only just beginning.