54 Business The EconomistFebruary 3rd 2018
2 cal Commissioning Group, near Manches-
ter. It crunches through the hospital’s
pseudonymised patient records, looking
for patterns that suggestthe emergence of
long-term diseases like diabetes and alert-
ing doctors if they are found. (In Septem-
ber Microsoft started a health-care division
in Cambridge, which will devise medical
algorithms of its own.)
None of this is straightforward. Health
records generated in conventional clinical
settings are immensely valuable—no
amount of smartwatch accelerometer data
is going to take the place of an MRIor an X-
ray. Such records are increasingly available
in digital form (see chart) but they are often
messy and hard to process. DeepMind has
had to spend months cleaning up data
flows from a British hospital, the Royal
Free, for instance, and has not yet delivered
anyAI-driven insights.
Apple’s approach is more focused on
hardware. It is working to make its devices
into trusted, secure channels through
which medical data can flow. Third parties,
rather than Apple itself, then build useful
health services on top. Putting patients’
health records on iPhones will make these
far more effective.
Millions of people around the world
have already joined medical studies using
this infrastructure, participating through
iPhones. Hints are emerging of the power
of the approach. An app called mPower,
built by Sage Bionetworks, a non-profit re-
search organisation, studies Parkinson’s
disease by getting iPhone users to perform
tasks and measuring the tremor in their
hands using the phone’s internal acceler-
ometer. The data so far suggest that Apple’s
platform may soon enable medics to spot
Parkinson’s digitally, over the internet, be-
fore it is symptomatic in a patient.
All this makes money for Apple by in-
creasing the perceived value of its devices,
says Anurag Gupta, a health-care analyst at
Gartner, an ITresearch firm. The firm also
hopes that clinicians and insurers will buy
and use Apple devices in their work.Algorithms v arrhythmia
Delivering medical services through
brand-new channels—the second route
into the industry—holds equal promise.
The new non-profit venture from Amazon,
Berkshire Hathaway and JPMorgan Chase
fits into this category. Details are scarce, but
the three firms might, for instance, use Am-
azon’s data-processing skills to build tools
to monitor and care for patients outside
hospitals and surgeries. Some speculate
that the firms might build an app that
makes booking a doctor’s appointment as
easy as reserving a table at a restaurant.
Whatever is built for their own employees,
it seems likely that the services will end up
being made available to the public, too.
Another recent innovation is Alpha-
bet’s Cityblock Health. Its mission to carefor low-income city-dwellers in their own
homes does not rely on existing health in-
frastructure at all. It plans to send its own
health-care professionals to the homes of
people needing care, with the visits paid
for by insurance, often Medicaid, the so-
cial-insurance system thatcoversAmeri-
ca’s poorest. Cityblock Health will trawl
data to spot where care is needed. It plans
to hire some 55 people over the next six
months, including data scientists, software
engineers and a lead physician, as well as a
team to interact directly with patients.
Smartphones and watches are chan-
nels for new services in their own right. In
this realm, Apple would become a direct
provider of actual care to patients. It holds
patents to turn its phones into full medical
devices, using a bundle of sensors aroundthe camera to let users measure their blood
pressure, body fat and heart function by
pressing a finger to the screen, for instance.
Other patent filings envisage putting sen-
sors into both phones and smartwatches
to collect electrocardiograms to monitor
heart health more precisely, and even do-
ing biometric monitoring through AirPods,
the firm’s wireless headphones.
Apple has also tinkered with sensors
which gauge stress or measure blood oxy-
genation, and is reportedly working on
ways to measure blood glucose through its
watch, helping it get a handle on diabetes.
Late last yearit said it would join up with
Stanford University to develop algorithms
to spot irregular patterns in heartbeat data
gathered by its watch. This research could
produce what is known as a “digital thera-
peutic”, which goes through a full regula-
tory approval process (see next story).
Verily’s eclectic range of ventures in-
cludes building new kinds of surgical ro-
bots in a joint venture with Johnson &
Johnson, a drug firm. It is also working on
cutlery containing self-stabilisation tech-
nology, to help those with tremors to eat.
The firm has two joint ventures to try to
tackle diabetes, one with Dexcom, a medi-
cal-device-maker, and another with Sa-
nofi, a drugs company. Verily also seeks to
organise health information so that it canbe queried for useful insights. Project Base-
line, a research study, will gather data from
10,000 people over four years to do this.
Facebook’s work in the field has been
discreet and focused on two areas—mental
health and clinical trials. In November the
firm said it had started to use AIto monitor
its users’ online behaviour for patterns
which indicate depression, and to reach
out in an effort to prevent suicide. The pho-
tos a user posts on Instagram may signal
depression, for example, depending on the
colours they contain, the times at which
they are posted and whether they show
faces. Google is trying something similar.
Facebook is also hoping to formalise
and monetise an activity which is already
common on its platform—namely, groups
of patientsdiscussingtheir symptoms.
One plan is to help drug companies recruit
people from these online gatherings for
clinical trials, and to manage groups of pa-
tients who sign up, presumably in return
for payment from the drug firms.This time may be different
It is worth remembering that the prospect
of technology firms transforming health
care hasbeen heralded in the past, only to
disappoint. Google started a health-re-
cords initiative in 2008, but shut itdown by
2011, citing poor adoption. Microsoft made
similar efforts with similarly low take-up.
Yet ten years on, the centrality of the smart-
phone, with its potential to give patients
access to their data whenever they want
and wherever they are, changes the game.
So too does the inexorable logic of the
data economy. Data sets that contain infor-
mation about human health are hugely
valuable. At a time when health-care bud-
gets around the world are stretched, payers
are desperate for insights that might enable
them to cut costs while maintaining quali-
ty. The more data the tech firms can handle,
the more they will learn about human
health, and the better the services they can
offer will become.
That raises some familiar concerns. Pri-
vacy is an obvious one: the tech world’s
mindset of “move fast and break things”
works less well when it comes to health
data. And the same competition issues that
dog Google’s search business and Face-
book’s social-networking service would
arise in health care, too, if a particular AI di-
agnosis platform, say, were to became
dominant. DeepMind’s work in Britain is
already on the radar of the EU’s competi-
tion watchdogs. Meanwhile, Amazon’s
partnership with Berkshire Hathaway and
JPMorgan Chase will lure in huge quanti-
ties of patients’ data, leading to continual
improvement of its services and, poten-
tially, fears about dominance.
Apple’s entry into the field offers some
answers to these worries. Its efforts in
health so far have been cautious. The only
patient data it processes now come from itsVital signsSource: Department of Health and Human ServicesUnited States, electronic-health-record adoption
by doctors in local practices, %0204060801002004 06 08 10 12 14 151