The Journal of mHealth Vol 2 Issue 4 (August) | Page 6

Digital Health: Making Sense of the Data

Digital Health: Making
Sense of the Data
By Dr Alexander Graham
Dr Alex Graham is a medical doctor by background, having trained in London before entering
the business world. He is currently a founding partner at AbedGraham, a research and strategy
consultancy which assists global IT corporates to navigate the clinical, organisational and commercial complexities of the UK’s National Health Service (NHS). He is also medical director
of EMEA for Imprivata.
One of the better analogies I have come
across regarding healthcare data is the
one with the faulty car engine. How can
we know so little about ourselves, when
even our car dashboard tells us immediately when something even small goes
wrong, such as a flat tyre or a broken
light? How often do otherwise healthy
people take their blood pressure or check
their blood sugar? I must admit I have
no idea what mine is, and I used to tell
people how to manage their conditions!
We are approaching an inflection point
in data gathering for healthcare. The
advent of sensors, internal and external, electronic health records and patient
portals means that the sheer number of
discrete data points will only rise exponentially. One of my favourite statistics
is that the world is producing as much
data every two days as it was in the whole
of time up to the year 2003, which is
truly staggering. Of course, healthcare
always lags behind most other industries
but the avalanche of numbers, diagnoses, treatments, interactions and histories
is coming and we are not prepared for it
at all as things stand currently.
Think about how much data is produced
from a single patient even now. Primary
care records, hospital admissions and
treatments, community care, day-on-day
and year-on-year. And what do we do with
all that data? For the most part, hold it in
user and institution specific silos without
any attempt to glean real benefit from it.
I remember in A&E, the sheer number
of patients who came from primary care,
the community or even just walk-ins that
you couldn’t find so much as an allergy
status about. Not only are we not realising
the benefits of data, we are causing harm
both clinically and financially.

4

August 2015

So data is the key then to a successful
healthcare system? No, data on its own
is almost entirely useless. The mere collection of numbers, statistics and records
serves almost no purpose if there is not
a concerted course of action to turn
that into tangible knowledge. Take a
patient’s allergy status, for example. Having the data as a stand-alone point on the
EHR or a drug chart is fine, but it only
becomes real knowledge when the prescription is made (or rather hopefully not
made) for that particular drug. Or when
we look at cohorts of patients and see
whether heart failure as a side effect is
higher or lower than the status quo. Or
how symptoms quantitatively respond
to the latest medication. That is the real
challenge here, making sense of the data
so that patients and healthcare professionals can actively change the way they
look at their health. Here are a couple of
things I think are important in data in
healthcare.

Structured and
Unstructured Data
The issue with most data is that it is not
in nice neat columns in the same software packages and the same for every
single patient. Most is concealed in the
sprawling mass of handwritten notes
and siloed departments. Data analytics
as much as possible requires the collection of structured data that is easier to
work with on a mass scale. The advent
of single-provider EHRs or even the
use of integration engines in the best
of breed (multiple disparate providers
doing different packages) model mean
that a push towards standardisation can
be made, although this must be considered as a viable entity in the design
and adoption of these upcoming tech-

nologies. Even if data is unstructured in
nature however, such as patient scans
or handwritten notes, the advent of
technologies such as machine learning
and natural language processing mean
that we can start to generate actionable
insights.

Incentivisation
The main question however, is not can
the technology handle it (even with
its limited use in healthcare, I trust
advanced technology to do what it says
on the label almost completely) but how
do we handle the human element of data
and knowledge? How do we get all the
individual stakeholders in the system to
jump on the bandwagon? Because the
first question I have for any revolutionary analytics tool or the like, is how are
you going to benefit everyone in the system? How will you convince the patient
to wear their sensor? How can you make
sure you don’t add to the workload of
a community nurse? How can you make
sure (cynical I know, but let’s not pretend this doesn’t matter), that a hospital doesn’t lose revenue or incur greater
costs? Those are the real questions
around what is ‘actionable’ data, because
if you have the greatest collection and
analytics machine in the world without
incentivised individuals, then data will
never become knowledge.
I have no doubt that analytics platforms
running on a constant stream of digital
data underneath will help to change the
way we practice medicine but until we
refine the catchment and transformation
into knowledge and understand how to
fit data into workers’ working patterns,
we will continue to lose out on the possible benefits. n