Module 5 Transcripts

Video 1: Module introduction [01:48]

In this module on the digital transformation of hospitals, we will take a look at the
following topics. We will first look at the core challenges that are faced by hospitals
and some of the approaches they're taking to overcome these challenges. Next, we
will get into looking at simulations and the powerful role that simulations have to play
in modelling out some of the transformations that are possible and doing these
changes in silico before doing it for real. We will then get into the detail of a specific
case study using a simulation technique.

Through the rest of this module, we'll also look at various productivity tools that are
improving clinical effectiveness in a hospital setting. We will also look at various
analytics tools that are now coming to market to benefit clinicians and patients alike.
We will consider topic of patient safety as well because whatever technology is
employed in a hospital setting to deliver care, needs to improve safety and outcomes
for the patients who get that healthcare service. But fundamental to this topic is
trying to understand, what is the future of the hospital, and does it necessarily have
to be a very large building? And we ask, what role does technology have to play in
creating the shape of the future hospital?

Video 2: The future hospital [05:00]

Hospitals they've been around for a long time. Some of them are dated back to the
fifth century BC when they were first recorded. There are place where people who
need a specialist help, support or service they go there. Hospitals are typically built
to last 50 years. Often, they last more than a 100 years.

Many renowned hospitals, if you visit them today, you will find buildings of different
vintage on the campus, they are dotted all around and they're connected with each
other with walkways, tunnels, bridges, footpaths and other means of convenience
between them. Completely new built modern hospitals are cathedrals of glass and
steel. The structural concept of a hospital has not changed for centuries. It is a place
where expensive, scarce, critical resources are concentrated to provide a quality and
continuity of care. You would want to go to a place where you have the right level of
expertise and equipment to deliver specialist care, like a cancer care centre, or a
specialist stroke treatment facility or a hospital that just performs organ transplants.
However, our world has changed over the centuries.

Today we live in a world where 70 per cent of our healthcare need is centred around
long-term conditions like diabetes or heart disease. In this changed way of living,
what should be the role of the modern hospital? What role does it have in supporting

Digital Transformation in Healthcare: Innovation, Strategies & Processes Page 1 of 29

people who have these long-term needs? And in a hyper-connected world, what
should be the type of service the hospital offers?

We need to take a closer look at how hospitals might work in the future. It can be
difficult to contemplate such futures when there's a multi-billion-dollar hospital right in
the middle of a modern healthcare system. So, what would happen if the only
hospital in town gets completely destroyed in an earthquake? What would people
build in its place? What role would that hospital perform? Well, this actually
happened in Canterbury, New Zealand. And we can learn from some of the
decisions that the citizens took there. When the citizens of Canterbury got together
to design their future hospital, they made some fundamental changes — some of
these are still in progress.

First of all, they wanted to see a shift from episodes of care to connected care
experiences to pathways of care. They wanted to see a shift away from centres of
excellence that operated in silos, to a system where they were all connected nodes
in the network. They wanted to have this big shift away from siloed organisations to
organisations that work together in a system, where the hospital as a centre of
excellence, worked with providers in the community and family physicians to provide
great all-around care for their citizens. Digital technology is merely a small enabler of
such a shift.

Hospitals have been using technologies all along to make incremental changes in
productivity and quality and safety. And they will continue to do so. This is a part of
the incremental innovation process. It's difficult not to get caught up in it, but one
must step back and take a look at the bigger vision. There is a much bigger price to
be won by using some of the advances in technology to think about what the role
and shape of the future hospital needs to be.

A truly digital hospital will work beyond the boundaries of its walls. Its services will
reach out directly into the community that it serves. It will work closely with the
primary care providers, the family physicians, the community care services and any
other services that patients require to deliver care in such a way that hospitalisation
is prevented to the extent possible. And in the event that a person needs to go to
hospital, no matter what is the time of day or night, they will always receive the best
care that is available to them.

Video 3: Challenges for hospitals [14:13]

What are the questions that keep hospital leaders awake all night? They will be
thinking, ‘Does the hospital provide safe care?’ In other words, they'll be thinking,
‘Where is the next lawsuit or negligence claim going to come from?’ They'll be
thinking, ‘Do we provide great care? Or what our patients and their families saying
about us? What are they saying about the hospital? Are we a respected brand?’

Digital Transformation in Healthcare: Innovation, Strategies & Processes Page 2 of 29

They'll be thinking, ‘How do we stay relevant? Roughly translated, that means ‘What
can we do to attract the best professionals to work here? 'Are we investing in the
latest equipment? Do we offer the latest treatments? Are we a great place to do
research? How do we become that respected hospital within this community?’

Hospitals often have to balance what might look like competing factors to achieve
their objectives. These factors need not compete, but sometimes it does seem so.
So, for example, if there's an efficiency drive in a hospital to try and reduce some
costs, is there a risk that the patient experience could be compromised? By reducing
staff, would you make patients wait longer? Even worse, by maybe cutting some
corners somewhere or pushing a process faster, would you compromise patient
safety? That is difficult to contemplate. Flipping it the other way, if you had a drive to
improve safety and to improve quality, would that drive up costs? Do you need more
staff? Do you need to invest in new equipment? That's something that's best thinking
about.

I find that tech companies will often run surveys and research on what are the
priorities of the healthcare CIO or the healthcare leader. And these surveys and
research papers tend to be rather self-serving because most of the time, they say,
‘Oh! you need to invest in telehealth and telemedicine to become more relevant as a
hospital or you need to invest in a data strategy or a data platform, you need more
workflow and more process automation or maybe he really need to be investing in
artificial intelligence (AI) and machine learning because this is the exciting thing
that's happening now in healthcare’And others will paint scare stories to say, ‘Well,
hospitals aren't doing enough around cyber security, there is a huge threat of the
leakage of patient data and compromising the hospital systems.’ Yes. These are all
important topics and certainly, something needs to be done about it. But are they
really transformative? Are they really the strategic focus of hospital leaders?

There is a concept called the quadruple bottom line that is claimed by more and
more hospital leaders as their strategic focus. And what is this quadruple bottom
line? The first is an improved patient experience to ensure that the patients are
always feeling safe and cared for when they are in that hospital. The second is to
deliver better patient outcomes and that means delivering a high quality and a safe
service that produces the health outcomes that the patient was expecting from their
interaction at the hospital. The third part of the bottom line is a great clinical
experience, and this means providing efficient processes for clinicians to do their job
more effectively in the hospital. And the final part of it is to provide financial balance
to manage the balance of investment and costs so that the hospital remains
financially viable to do the job that it needs to do. But the dimensions of the
quadruple bottom line are not necessarily the priorities that guide hospital managers
at the next level down in their day-to-day operational decisions.

And we will consider here the slightly different but similar priorities of hospital
managers in a public hospital vs those in a private hospital. In a public hospital, the
operating managers will be concerned about reducing waiting time or to reduce the

Digital Transformation in Healthcare: Innovation, Strategies & Processes Page 3 of 29

length of stay once a patient has been admitted to hospital. When a patient has
completed their treatment, they will be concerned with discharging the patient
efficiently. And once they have been discharged, they would want to see that the
patient is not then readmitted for the same condition because that is an indication
that they were discharged too early or the treatment was not successful. While the
patient is in hospital, they would want to do their best toward infections or to reduce
or eliminate any other avoidable harm that the patient might come to in a hospital.
And finally, they would want to optimise the utilisation of the hospital in terms of
making sure that all the rooms are occupied, but not too full that all of the assets
such as diagnostics, scanning surgery are all optimally used.

In private healthcare, on the other hand, the priorities are somewhat similar, but the
emphasis can be different. Firstly, they would want to be the hospital of choice — if
someone needs private healthcare, they should think of their hospital first and make
that their destination. Once the patient is in the hospital, they want to provide a great
patient experience so that the prospect of receiving healthcare in their hospital
doesn't seem so unpleasant. They want to provide great facilities to attract the best
professionals, so that means having the latest equipment, having a great offices,
having a great work environment. They would want to provide all the required
treatments under one roof so that once the patient comes to them for treatment, they
wouldn't need to go anywhere else, they wouldn't want to go anywhere else. And by
providing all of those treatments in one roof, they need to balance that really
carefully to ensure that all the assets that they've invested in are used to the highest
extent possible.

But these two are just lists. What's most important in any such list? Let's take the
public hospitals as an example and I'm just picking out one at random here. Let's
say, make an argument for why infection control needs to be the most important
priority of a hospital manager. If a hospital is managed in such a way to minimise or
even eliminate the spread of infections in it, it can have effects on other priorities as
well. If there is no spread of infections, then firstly, you're taking out one cause of
avoidable harm. By taking out that cause of avoidable harm, you ensure that a
person doesn't stay in hospital any longer than they need to. So, you're reducing
length of stay in hospital. By reducing the complication and infections, you're also
improving the patient experience. So, you can see how by hitting one lever of
control, you can impact many other priorities that are relevant for hospital managers.

Hospitals are always improving and they consider various different approaches to
transforming themselves. But technology doesn't come to the forefront in many of
these decisions. So, let's look at some of the transformation options that hospitals
consider.

The first of these is consolidation through mergers or acquisitions. And there are
certain benefits that the hospitals can achieve through such transformations. The
first is they can consolidate the back-office processes. So, if two or more hospitals
merge, they can combine their HR, payroll, procurement and get a lot of efficiencies

Digital Transformation in Healthcare: Innovation, Strategies & Processes Page 4 of 29

out of combining those operations and save some money. On another dimension,
they could start pulling some of their clinical resources. So, one hospital could
become a cancer specialist centre by getting all of the cancer specialists to work at
that location. The other hospital could become a cardiology specialist centre by
getting all of the cardiologists to work at that location. So, those synergies can be
built up through a process of consolidation.

The other approach that hospitals could take is vertical integration. In this situation, a
hospital or a hospital group could start providing primary care services. By providing
the first front door to looking after people in a primary care setting, they can pick up
patients early who might just need to receive acute care in the hospital. This could
benefit patients by early diagnosis of certain cancers if picked up early could save
lives.

Equally, they could be quite valuable in picking up early diagnosis of certain

cardiovascular conditions and ward off most serious conditions like a heart attack or
a stroke. But these strategies also worked to funnel demand into the hospital and
improve the utilisation of the hospital itself. But overall, it can improve the economics
of providing healthcare within that system. Once a person has come to hospital and
received treatment, they are then discharged and go back home. The hospital could
then invest into providing home care services. This would ensure that patients who
are being discharged go home with a package of care, this would help with the care
of the patient as they are recovering and they complete the recovery at home. It will
help reduce readmissions into hospital and may even promote an earlier discharge
and free up hospital capacity and beds by creating the right package of care through
a home care solution.

And finally, hospitals can also invest into networked services and support by
providing some of the specialist advice through other doctors in the network or even
to patients through technology platforms like telemedicine and remote patient
monitoring. The final transformation option is outsourcing. And here the goal is to
take our processes from the hospital and pass it on to specialist providers. Some of
the processes, just back-office processes like payroll and procurement and others
might be niche processes like pathology, radiology or genetic testing.

In all of these cases, technology can be a key enabler. So, in mergers and
acquisitions, for example, if the merging hospitals have similar systems for either the
business processes, these are enterprise resource planning systems that include
finance, HR, procurement and so on or if they have the same electronic medical
record systems, the EMRs. Having shared systems can speed up the process of
integration and reduce a lot of the risk and the cost of the integration and merger of
these two or more hospitals that are coming together. There is good precedence for
this economy from other industries. We've seen in the energy industry or in the
pharmaceutical industry, how the adoption of common systems across the industry
triggered off a process of consolidation. So, it begs the question, would the rollout of

Digital Transformation in Healthcare: Innovation, Strategies & Processes Page 5 of 29

some of the global EMR systems trigger off a process of consolidation amongst
hospital groups? We shall wait and see about that.

Vertical integration, on the other hand, requires a different set of technologies. There
are to my knowledge no EMR systems that cover all of the healthcare settings from
hospitals to primary care to community and mental health. So, if an organisation is
pursuing a vertical integration strategy, then it needs a different set of technologies.
The key among these is interoperability frameworks. This could be an interoperability
software that sits as a layer between the systems of the various organisations that
are merging together.

And what this software would do is ensure that the data that's emerging from each of
these systems is easily shareable across the organisation and that common services
can be exposed in an equivalent way to everyone who needs them. The merging
organisations would also need a workflow system to ensure that task that originate in
a hospital can then be directed into a primary care setting or a community setting to
organise packages of care, reviews and referrals of a particular patient or to
otherwise organise complex activities that cannot be driven from any single system.

Video 4: Innovation management in hospitals [06:07]

For hospitals that want to be leaders in innovation, creating a dedicated innovation

management function is critical to their success. Leading hospital institutions in the
US have already done quite well at this—they create this dedicated innovation
management capability to support their organisation. It is not that this is a
department whose responsibility is innovation. It's more a set of specialists to
understand how the innovation process works and they can work with every other
group in that institution to design, test, evaluate and deploy innovations that can
have impact at scale and to roll those out systematically through the institution.

When the innovation management functions consider various innovation

opportunities, they usually analyse it on two dimensions. How relevant that
opportunity is to the core business process? And second, what is the strength of
relationships that is required to make that innovation successful? If the requirement
of strength of relationships is high and the relevance of the innovation to the core
business is high, in those situations, they will look at people within the organisations
and team within the organisation to work on that innovation because in that situation,
employees, they have the best understanding of the current business and they have
the best strength of relationship within themselves.

On the other hand, if the innovation is core to the business but it doesn't require a
strong relationship to make that viable or possible, then it becomes possible, or
indeed beneficial, to look outside the walls of the organisation and perhaps, you
know, run a tender or run a competition to invite people from around the country, or
perhaps even around the world, to say, ‘Hey, we've got this problem, we need an

Digital Transformation in Healthcare: Innovation, Strategies & Processes Page 6 of 29

innovation that helps solve it. Do you have something?’ And then one can go through
a process of evaluating those various options that come back to work out what is a
suitable innovation to implement for that institution?

There are some innovations that require a very strong relationship to implement, but
it might not sit very core to the business. So, for example, if a new drug needs to be
discovered, the hospital might not be the best place to do it but it may have the
patients that need that drug or that therapeutic. In these situations, it can work with
various industry partners, where they work in an area that's not core to their
business, but the strength of the relationship is strong. Then the hospital teams can
work with their business partners to co-design, co-produce, test this new innovation
and then work together in its implementation.

The final scenario is where the relevant innovations might not even sit core to the
business and do not require a great strength of relationship to implement. In these
situations, the innovation management function can look far outside the organisation
and go on a scouting mission to look at. ‘Hey, what are the interesting things
happening, perhaps in Silicon Valley or in China or in South Korea, that might be
relevant to our institution back home?’ I call this process technology scouting or
innovation tourism and it can often bring back valuable insights for implementation
back home.

Large teaching hospitals already have an established research base that's producing
outstanding medical and surgical innovations. Their leaders have now turned their
attention to how digital technology can drive that transformation. Imperial College is
one such institution that has invested in a number of innovation assets. It has, for
example, the Translation and Innovation Hub or the I-HUB for short. It has the
Enterprise Lab, which is an incubation unit for start-ups that have emerged out of the
Imperial campus. There is the globally acclaimed Institute of Global Health
Innovation, that is a thought leader in the space and is working on some outstanding
aspects of technological research. And the university has a number of other
excellent research assets, including close relationship with MedCity, which is a
research and technology park for the medical technology sector.

Technology can help hospital leaders achieve the quadruple bottom line but
implementing a technology-enabled transformation requires a huge change
management effort. And in my experience, I'm not sure if hospital executives
sufficiently appreciate the need for investment in change management. I have found
that a change management effort frequently requires three to five times the budget of
the cost of the technology itself. So, it's an effort that is important and cannot be
ignored. If you find yourself in a situation where you're part of a technology-led
transformation, especially in a hospital setting, do bear this rule of thumb in mind and
ensure that hospital leaders and management and at every level invest sufficiently in
change management to assure the success of the technology transformation.

Digital Transformation in Healthcare: Innovation, Strategies & Processes Page 7 of 29

Video 5: Introduction to simulations [07:07]

What are simulations and why are they important to healthcare? Well, simulations
have been used for a long time through history for taking complex decisions, often
under a lot of uncertainty. They've also been used for training—military training is a
form of simulation. Ever since the introduction of computers, digital simulations have
become a powerful, rapid and a really cost-effective way to run simulations to help
with this process.

Simulations take input data about a particular situation, some of the characteristics
and behaviours of the system and help play out any potential future scenarios that
might result from it. So, decision-makers can use simulation techniques to
understand every complex interplay of factors that may relate to a business decision
or a problem, work through the scenarios and use the insights that they gain to take
some decisions about the future. What's more? As the future starts playing out, they
can take some of the data that's emerging from the reality, put that back into
simulation and then recalibrate the decisions that they might need to make.

There are many different types of simulation techniques and tools. So, the tool you
choose really depends on the type of decision you have at hand and so it is
important to pick the right tool. Many of you may have already used spreadsheets.
These are quite useful simulation and forecasting tools where you can plug-in
assumptions and formulate, do a bit of what-if analysis. But have you ever tried to
create a feedback loop in a spreadsheet? I don't think that's possible. And a lot of
problems in business and nature in society have feedback loops. You need specialist
software and types of models to create those scenarios.

Discrete event simulation (DES) is one such technique. It is useful for problems that
have an optimisation goal. So, for example, you may have a question, how do you
help make this department run more efficiently? How do you improve the productivity
in this clinic or this hospital? How do you optimise this process flow for managing
diabetes or heart failure? It's good for those sorts of processes that operate over a
limited time frame and you want to optimise how resources are allocated, what are
the shift patterns and how workflows from one end of the process to the other.

You would look to system dynamic simulation for much longer-term problems.
Situations like, if our population changes like this over the next 10, 20, 30 years,
what sort of hospital do we need to build today to serve our needs in 30 to 50 years
time? What's going to be the nature of health pressures to try and create those very
long-term scenarios that affect populations that affect the resources of a society?
And then get some insight into the decisions that need to be taken today that will still
be relevant after two or three decades.

Simulations are a form of predictive analysis. So, when a simulation model is being
developed, it's developers will factor in some assumptions about how the current
system or you know the hospital department or clinic or that aspect of society

Digital Transformation in Healthcare: Innovation, Strategies & Processes Page 8 of 29

behaves. They will take some past data to calibrate that model to ensure that the
behaviour of that model reflects some of the behaviour of the problem that we've
seen in the past, that helps ensure that it's been set upright. They will then try and
model the impact of any changes or shocks to that system. What would happen if,
for example, there was extreme weather or there was another pandemic or in a
positive sense, if we introduce this innovation or if we made this transformation, what
would be the impact of their change on that whole system?

The predictive power of the simulation model depends on a variety of factors. First of
those is what is the complexity or the scope of that model that's been developed or
the problem that it represents. The second factor is some of the quality of the input
data. You never have perfect data going into these complex problems. There are
often gaps that need to be filled and assumptions that need to be made. So, there is
a degree of variability and uncertainty about how that simulation is going to
accurately predict the future. And the third aspect of the predictability of a simulation
is what control do you really have on the big levers that drive change in that
simulation to create the future scenarios?

We live in an uncertain world. A lot of the aspects that affect our future are outside
our control and those factors would also affect what predictions the simulation is able
to come up with. We have a separate video on predictive and prescriptive analytics
where we look at how these tools are effective problem-solving or productivity tools.
But in this video, we'll specifically have simulations and how it can help us solve
either short-term or long-term problems.

We are going to look at discrete event simulation (DES) in particular because this is
quite powerful in solving problems in healthcare. Many problems of quality and
safety and experience in healthcare situation tend to be issues related to the flow of
people, material and activity through the healthcare system. By creating a discrete
event simulation (DES) model of the flow, it is possible to develop quite good insights
into where are the bottlenecks? Where is the problem? What is it that is holding up
the performance of that pathway?

Models have a great benefit in problem-solving. They help the team members who
are participating in that problem-solving exercise, focus on the facts, focus on the
evidence, they can take the opinions and put it into assumptions and have a healthy
debate about solving the problem, developing different scenarios and evaluating
them.

Video 6: Introduction to Flow Coaching [04:26]

A discrete event simulation (DES) modelling software or a tool in itself will not help
you create really innovative solutions to a flow-based problem in healthcare. And so,
I'm going to introduce you to a different problem-solving technique that has been
used quite a lot in healthcare, here in England to solve flow-based problems. It can

Digital Transformation in Healthcare: Innovation, Strategies & Processes Page 9 of 29

help you come up with creative solutions and then discrete event simulation (DES)
can be applied to test out whether those creative solutions are going to work or not.
Note, you could use these techniques in any other industry, it doesn't have to be
exclusive to healthcare.

The Flow Coaching Academy started at Sheffield Teaching Hospital, and it has been
doing a lot of work with hospitals in England and now other non-hospital healthcare
organisations in solving flow-based problems. It relies on a number of fundamental
principles for its success. The first is, it is based on technical quality improvement
methods or QI methods. The second is our huge dependence, in fact, the foundation
is laid on statistical process control (SPC) and the understanding of how processes
actually perform. The third is, it depends on strong coaching skills and the ability to
have active listening. There is a great emphasis on listening, supporting and
providing feedback. There is a lot of training that's given to facilitation methods and
to time management to ensure that the whole flow coaching process runs smoothly
and is as effective as possible.

Flow coaching sessions take place in real or virtual big rooms. So, what happens in
these big rooms? First of all, they would assemble a multidisciplinary team (MDT)
that can help address a problem. So, there's bringing together of a whole different
set of experiences and skills. Secondly, they suspend all hierarchy. So, people can
have a strong debate with someone who might be their boss, but the hierarchy does
not matter. The third aspect is in preparation for solving this problem, they will gather
all of the background materials and put it up on the walls. So, all of the information
related to the problem is there for everyone to see, everyone to discuss, debate and
then to act upon it. So, this is this creates that environment in which the whole group
can work together to solve the problem. Building up the capacity to keep this process
going involves a lot of co-coaching. What this means is that in the flow coaching
process, new coaches are developed who are supported by more experienced
coaches and that creates the capacity not just in the organisation but in the wider
flow coaching network and community to improve this quality improvement process.

So, now let's assume you're in this big room. The team is all there. You work through
this flow coaching process and you've come up with two or three different options to
help improve the process. How would you evaluate what is the best option? Now
traditionally, people have used pros and cons analysis or a cost-benefit analysis.
You could use attention directing tools like you know plus points, minus points,
interesting points of each option or you could assess the strengths, weaknesses,
opportunities and threats doing the SWOT analysis. But consider this; you could use
discrete event simulation (DES) to test out the various options

Video 7: Using discrete event simulation (DES) in healthcare [21:29]

Hi, I'm Tom Stephenson. I'm the Director of Services and Strategic Partnerships at
Simul8. And Simul8 is a discrete event simulation (DES) software company. We’re

Digital Transformation in Healthcare: Innovation, Strategies & Processes Page 10 of 29

based in Glasgow in the UK and what we do is we build pathways in all different
sectors. So, we have manufacturing flows, supply chain and logistics, oil and gas all
these different areas, but for me, one of my personal priorities or preferences is
building pathways in healthcare and that could be pathways looking at flows through
hospital or it could be looking at the health of the population, it could be looking at
the impact of a new medical device or a new drug.

And what is Simul8? So, we are a software that make discrete event simulation
(DES). What you should see is a picture here of board game, essentially a big board
game and this is from 1994. Where our found was working for British Leyland, and
they are company that make cars, especially the mini, they may have heard of. And
Mark, our founder at the time, he was using a lot of complex formula to understand
how to change the manufacturing line. And while this was good and while it was
accurate, he found that he was struggling to get buy-in, sometimes from senior
management who may not have the full understanding of it. So, he went about
building this big board game and what this board game is a virtual representation of
the factory of building the cars. There's little poker chips that move around, just like
the cars move around. And people could sit around and see this with stopwatches,
looking at the flow of events.

And what we can find by building or having this flow in front of us, is that it adds a lot
to the understanding of how the system works—it can help you have good
conversations and actually, our founder who invented this was then taking it to a lot
of different factories to use it because they found it was so effective. But it's not really
too good having this big board game where we want to look at all different types of
flow and improving it. So, Mark went away and he built software that could
essentially do the same thing. And this is where simul8 started. So, we can build
outflows in the software by dragging objects on the screen to represent different
steps in the process.

And what I'm going to do is actually show you a simulation, a healthcare example, so
that you can understand how this works easily in a healthcare setting. So, here we
have a simulation of an emergency department. We've got patients arriving either on
foot or via an ambulance and they're going to come into the waiting room, they're
going to be treated in the cubicle and then they're going to go to a bed potentially if
they need it. We've set this up as a demonstration to represent the flow in an
emergency department and we want to configure this to match the demand that we
have from patients. As the simulation runs, we will collect metrics, such as how
utilised the different beds are, how used different staff are and how full our waiting
room is. And this is going to help us to see how well the simulation is performing. If
patients wait too long, they may go to the hospitals. Now, I'm running this for one
week and you can see the clock moving, but actually, I can run this much quicker to
get to my simulation results.

There's a lot of results here—we've got utilisation results, we've got cubicle blockage
and we've got waiting times. The one that I want to focus on here, is the waiting

Digital Transformation in Healthcare: Innovation, Strategies & Processes Page 11 of 29

room time. So, the percentage that have queued less than four hours is only 11 per
cent to start with. And we want to get this much, much higher. So why is it so low? If
we look at the reason for a cubicle blockage, we can see that probably the most
likely reason is that ER beds are causing the block because these cause blocks 34
per cent of the time. So, this is where I would change the system and see what the
results are going to be.

So, I'm going to come into my settings and I'm going to change the amount of beds
I've got—from two to four—this caused the biggest block. So, now I would expect if I
re-run the simulation with everything the same, that block should be reduced and
therefore my system should perform better. Because this is a virtual representation
of a real system, it should be that if this change has a positive effect in the virtual
system, we have good evidence that the change will also have the same positive
effect in real life area. I'll run this simulation through much quicker. And while we can
see that this number has improved, its maybe not improved by as much as we would
hope. We can see that ER beds are no longer causing a block. But it appears that
now much more patients have been blocked because we don't have enough doctors
and we can see that by the doctors' utilisation being almost a 100 per cent. So, we
haven't actually alleviated the problem here, we've just moved the problem
elsewhere.

So, now I can see that also, if I'm going to increase the number of beds, I would
need to increase the number of staff in line with this. And we can have a lot of
different ways of looking at staffing or any settings in the model. In this case, we
have got some different shifts. So, I'm just going to increase the amount of doctors
we have on shift and now I'm going to re-run the simulation and run it quickly this
time so we can get our results really quickly. This is a marked improvement. We can
see that we're now at 80 per cent and by making this change, the system is working
much, much better. We can see we have actually pushed the nurses’ utilisation of
quite high now.

So, if we were to really modify and optimise this system, we may now want to also
change the nurse shift and add one more nurse onto one of the shifts here. Okay, if
we re-run the simulation one more time, we can now see, we've hit 95 per cent and
we'd be happy with this. This is how you can use a simulation—you can change
different settings and see what the impact is likely to be, in the real system.
Hopefully, that's given you a good introduction to simulation and how simulations
look and how they can work.

But why is discrete event simulation (DES) a useful technique? First of all, when
we're looking at any system, there's going to be a lot of variability in how that works
and especially in healthcare, we're going to see patients have much varying glimpse
of stay, we're going to see that treatment pathways are going to be different, and we
might not even change anything for them to be different, but that variability has a
really big impact on flow. And especially when we're looking at waiting times, we can
actually see there's a distribution on the screen, which shows the pattern that we see

Digital Transformation in Healthcare: Innovation, Strategies & Processes Page 12 of 29

in a lot of healthcare facilities, whereby broad majority of patients are only staying for
a certain amount of time. But we have this big tale of patients who may wait much
longer or maybe in the system much, much longer. And if we don't capture that
variability, we're not going to get a good sense overall of how the system works
because they can really take up a lot of the capacity we have. We try and build in
other elements of variability as well. So, we look at seasonal trends, we can look at
weekly patterns as well to see how we might need to change processes on a week-
by-week or a month-by-month basis. When we actually run a simulation, we run it
hundreds of times as well and we'll take confidence interval. So, what that means is,
we won't just report and say our waiting time is 20 minutes. We'll say that we are 95
per cent confident our waiting time is going to be between, say, 18 minutes and 22
minutes—and this allows you to capture the risk that we will see in making real-life
changes.

The next thing that's really good about simulation is that visual element that you have
seen, so being able to see the flow helps you to get engagement first of all. So, if I
go into a meeting with a ten-page report, I often don't get as much engagement as if
I go in showing a simulation. What that leads to, is better communication—you can
ask better questions and you can get a better understanding by having those
discussions with the people involved.

The other thing is, we can understand the complexities here, especially when we're
very used to working in one particular area of a hospital or healthcare system. It's
good to be able to see how those pieces fit together in a visual way and all of that
should lead to better solutions.

So, the final advantage of simulations we can see the dependency that we have, so
all the individual steps are linked to one another, just like they are in real-life. So, we
can see, for example, what if there aren't recovery beds available before a surgery
starts? Are we going to proceed with that surgery, or might we consider cancelling it?
What if there aren't enough recovery beds in the community and people aren't
medically fit to leave the hospital? But there's nowhere for them to go. So we could
see how that could cause a build-up further back down and in a similar way in our
emergency department. Is that going to work effectively if there aren't beds for those
people who need beds to go to? And what we see is that when we get a blockage in
one area, it's going to feedback all the way down the system. So, that's where having
that interconnected system, it's going to be really helpful.

So, now you've seen an example of how this works. I'd like to get into a real example
of a previous project that we've done, so you can see how this technique would get
utilised. We're going to look at a sepsis example and sepsis is effectively an
inflammatory response to an infection and it's going to progress. So, we can come in
with what we think is a normal infection and that can quickly progress to sepsis. That
sepsis can become severe and eventually we could go into septic shock. So, it can
be easily treated. But it's quite difficult to diagnose and actually once we have sepsis,
it has a really high mortality rate of around 40 per cent. And in this simulation, we've

Digital Transformation in Healthcare: Innovation, Strategies & Processes Page 13 of 29

used a baseline of 37 per cent. And what this simulation is looking at is to assess the
impact of having early intervention.

So, we're going to use a new technology which improves early diagnosis, just using
a quick test. If we can diagnose better, then we can get people on an appropriate
course of treatment faster and also, it means that we can just treat them based on
whether they do just have a local infection or whether they are at a different stage of
sepsis. And ultimately by doing that, we should reduce deaths.

So, we started off by mapping out the pathway. So, we mapped out the steps that
patients move through. You can see that we haven't quite changed the visuals of this
so that it looks completely like a hospital. But you should recognise that the steps are
different hospital areas. So, we've got patients arriving, then you are going to have a
bit of a wait for diagnostics and things like that, they're going to have some tests
done and then once the results of those tests come back, they are going to go to an
appropriate area of treatment. It could be that they've been misdiagnosed after that
initial testing. So, they might have to come back around this loop and go to different
treatment areas as well. With the introduction of this new device, the area that we're
looking at is that diagnostic section. We're trying to get a result quickly about what
stage of sepsis they are in, if they are in a stage of sepsis, so that we can get them
to treatment much quicker.

So, we've mapped out all the steps here. But then it's important that we add flow
information so that the flow represents reality. We've got different information about
the patients. So, this could be things like what age are they, it could be things like
what gender are they, we can have all different information but also we need to have
information about how sepsis works. So, how quickly does it develop in people? And
we've got here some stages of sepsis that patients will go through based on the time
until they're treated. If we can get that time down, hopefully it will prevent sepsis
getting worse and it will give sepsis less chance to get worse.

So, after we've done all that, we do a lot of runs of the simulation and we can access
the results of this. Now, I've just got a couple of results. One is the baseline. So, we
always run the baseline and then I compare that to what if we use this new device
that tests quicker. And you can see that it's really good because it allows patients to
get treatment quicker, allows them to actually be in treatment less time as well
because they're getting the appropriate treatment. It is going to bring down the cost
of treating them. And importantly, we can actually see the deaths have decreased
here. So, we've lost 5 per cent of our deaths by introducing this device.

Secondly, because we've got the flow mapped out, we can look at how the capacity
changes, as well, could further improve the flow. So, we've looked at actually if we
can get more people to the right treatment pathway quicker, we need to have that
capacity there to be able to treat them and with added nurses as well as having the
new device. And here we can see even further improvements. We can see that our

Digital Transformation in Healthcare: Innovation, Strategies & Processes Page 14 of 29

deaths have now reduced to 9 per cent, for example. So, by doing this and relatively
testing different things, we can see a numerical value against our results.

So, when we are building simulations, there are a few key steps that we want to go
through to ensure we get results. First of all, we really want to define the problem.
What are we testing? What do we want to see the impact of? If we define this clearly,
we can look at what results do I need to see if the simulation is effective and we can
prevent wasted build time. Also, we want to involve a team on a simulation project. If
it's just one person building the simulation by themselves, it might not have as much
impact as if people are involved, especially people who know how the system works
and how the flow works.

We're going to want to build the simulation iteratively, we don't expect to get a
system that's perfectly reflective of reality first time. We would rather build things
quickly and take learning as we go. Also, we want to test lots of what-ifs. So, it's not
just about building a nice-looking model. It's all about running different scenarios and
comparing result. And I'm going to go into a final example where I'm just going to
show this and show the different scenarios that you maybe outrun. In this simulation,
the aim is to reduce our waiting lists for surgeries.

We can see we start off with a waiting list of 150 patients across three different
specialities. So, there's three different specialities that are using these surgical
rooms for procedures. Also, we have emergency arrivals coming in and those
emergency arrivals are going to go to our emergency beds. If there isn't an
emergency bed available, because two emergency patients come at once, they are
going to use one of our general beds, which will mean one of our electives who is
waiting for a procedure, will get cancelled.

The idea is that we can hit our target waiting list of only a 140 patients after 12
weeks. So, we need to bring the waiting list down. But what we can see, looking at
the result that changes as the simulation runs, is that when we look at all three
specialities combined, the waiting list is actually increasing. So, if I run this through a
little bit quicker, we're going to see that by the end of week 12, our waiting list has
risen significantly. And the idea of this simulation is that we're going to try different
tactics to reduce this waiting list and decide which one is going to be best because
there will be more than one way that could work.

The first option is to simply add capacity. If I add surgical beds, I'm going to be able
to do more surgical procedures which should result in the waiting list being reduced.
When I run this simulation, we can see that's made quite a marked improvement.
Actually, by week nine, we can see we've cleared our whole waiting list. And we can
see that that's gone down completely to zero. This is going to be really good, but it
might not be feasible to add two surgical rooms. It's going to be expensive, it will
mean we have to staff it. So, we can try different options here. I'm going to reset this
simulation and I am going to take away one of my extra beds. But also what I can do
is de-prioritise a certain speciality. So, it might be that speciality A and speciality B

Digital Transformation in Healthcare: Innovation, Strategies & Processes Page 15 of 29

have conditions that need to be treated quickly. It might be that they could save lives
or patients who are very serious could be in these categories, whereas it might not
matter so much if the waiting list is a bit longer for these patients. So, it's a different
configuration. It might not be as expensive, but it's going to mean potentially a worse
result for these patients.

As I run this simulation, we should start to see a separation in the waiting list. So, we
can see that we are reducing the waiting list for speciality A and speciality B. But
speciality C, where the line is the same colour as the patient, is going to have to
grow and at week four that we're on now, we can see this continues to grow. The
good thing is that the waiting list general trend is downwards even though we've only
added one surgical unit.

Let's run this faster so we can see the results at the end. So, the good news is we've
reduced our waiting list. It's down to 106, which is better than our target. And even
though it's not as low as if we added two surgical beds, it has gone down. But also,
these two groups that we consider quite serious, if we look at the more detailed
results by clicking into the spreadsheet, we can see that we cleared the waiting list in
31 days. So, just over four weeks. And that's compared to nine weeks if we hadn't
have de-prioritised but having an extra bed. We could try a lot of different things with
this as well—we could say, ‘Rather than de-prioritising, why don't we treat a certain
speciality group elsewhere?’ And this is certainly something that's happening in the
UK at the moment.

People are going to private healthcare facilities or newly built facilities as a means of
taking them away from this group and being treated elsewhere. This is how you can
use a simulation to understand which option might be best. So, you can see how you
could do both of these solutions but it's all about saying which one is going to be
best. It might not just be about de-prioritising. It might be that actually, we could send
certain patients to different treatment facilities in the UK a lot were using private
treatment facilities rather than the NHS. And those types of tactics can all be tested
and we can see what the impact is going to be.

Video 8: Productivity tools [20:29]

Hospitals place great emphasis on improving productivity. An improvement in

productivity means that a hospital can treat the same number of patients, assuming
that other constraints of staff, space and equipment remain the same. By improving
productivity, when staff being are able to treat more patients, it means that
healthcare professionals have an opportunity to improve their skills even further
through their experience—the ability to treat more patients also means that the
hospital has an opportunity to earn more revenue. It is no surprise, therefore, that
most of the productivity improvements that are authorised for implementation in a
hospital setting tend to be related to the improvement of productivity, as much as 80
per cent of the time.

Digital Transformation in Healthcare: Innovation, Strategies & Processes Page 16 of 29

Hospital productivity tools can broadly be classified as follows: there are tools that
improve record-keeping, there are others that improve business process workflow,
there are tools that improve clinical productivity and throughput, there are others that
improve safety, there are some tools that improve the utilisation of high-value assets.
Other tools improve the utilisation of people and then you have a class of tools that
help reduce waste and ensure that any of the productive gains from the other tools
are not lost.

The first category of tools that we consider is those that improve record-keeping. The
most basic amongst these are patient administration systems or just called PAS. The
patient administration systems (PAS) maintain a database of all of the patients who
have been treated by a hospital. They contain some basic information, such as the
patient's name, address, date of birth, gender, etc. And they catalogue all of the
treatments that the patient may have received and serve the purpose of doing
appointment scheduling and billing in the main. This is where hospital system
started.

As these PASs became more mature, various departmental systems are added on to
them and they became more comprehensive electronic medical record (EMR)
systems. Now, although there do exist some families of comprehensive inter-
departmental and medical records systems, most hospitals around the world tend to
have very siloed departmental systems. When an audit was done some years ago of
the number of different systems in a hospital, that hold patient records, patient
identifiable data—these are fragments of medical records. The audit found that there
was an average of 80 such systems in hospitals in England.

Some hospitals had as many as 500 different systems that held patient data. So, you
can appreciate how fragmented an electronic medical record (EMR) can get. Within
the same category, you also have electronic health records (EHR). Now, these take
the scope of electronic medical records (EMR) a little beyond the hospital and start
incorporating other datasets, particularly from primary care. These records are
become more informative about all of the care needs of an individual and incorporate
a level of sophistication that is taking EMRs to the next level.

You can take this even further in some of the local health and care records (LHCR)
that are being implemented in countries like England. In this situation, you've taken
the scope of the electronic health record (EHR) and added social care and
community care data into it as well, which traditionally fell outside of the scope of
either hospital care or primary care. By combining all of this public service data that
is related to the health and well-being of an individual, it enables public services to
get a better view of an individual's needs and then to be able to organise the health
and care services around the needs of each specific individual.

Personal health records or PHRs were created to help individuals take better control
of their own data because from a patient perspective, they're going to see more than

Digital Transformation in Healthcare: Innovation, Strategies & Processes Page 17 of 29

one doctor in more than one hospital for their specific needs, especially if they have
a long-term condition. It can therefore become quite difficult and complicated for
them to keep track of all of the medical data. In the past, it used to be done through a
collection of papers and files and CDs or DVDs. But soon there emerged software
that enables patients to catalogue all of their data into one place. There are now
hundreds of personal health record (PHR) apps. Not all of them are very good or
comprehensive or even fit for purpose, the best of them tend to be very niche-
oriented, there are a few that are more generic. These are valuable sources of
information for the patient to control their own information and to share it and have
meaningful conversations with their doctors, no matter which healthcare setting they
go into. This area has also attracted a lot of interest from the big health tech
companies and they still trying very hard to get into this space. There has always
been slight issue of privacy and trust of big tech and that is perhaps something that's
hindering greater uptake of PHRs.

One cannot talk about any of these types of health records without accepting, some
of the challenges of dealing with health data. The majority of health data still remains
unstructured in the form of clinical notes or messages or documents. Extracting
structure and meaning out of all of this unstructured data still remains difficult and
has been a challenge for health informaticians for the past three decades. Many of
the electronic medical record (EMR) systems as well place a huge burden on doctors
to maintain them. So, far from being a productivity tool, they have become an
additional workload. There's a lot to be done in this space and it should be watched
with careful interest.

The next category of productivity tools are business workflow tools. Some of the
more mature EMR and EHR systems do have some of their own workflow tools.
These are used to trigger off orders for pathology tests or imaging, they can also be
used to generate electronic prescriptions or referrals to other specialists or other
institutions for additional or further treatment. So, there is a catalogue of such
productivity tools and workflows within EMR systems. But as we have seen, a
hospital can have more than one EMR system. What do you do when you have to
trigger a workflow between one or more of these disparate systems? This is when
specialist workflow tools are required.

It is in this space that interoperability and robotic process automation (RPA) tools
have a stellar role to play. So, an interoperability piece of software sits as a
middleware between the clinician and various other EMR systems that might exist
within their hospital. So, if you imagine a situation where a clinician is meeting a
patient who has had several treatments in the hospital, they now would not need to
log in to different systems to find out what happened. If there's a good interoperability
piece of software there on that one screen they'll be able to see at a glance, the
details of the patient's interaction in every single department over a period of time
and from that same screen they will be able to then drill down and see additional
details that are pulled-up live from the underlying systems. So, interoperability

Digital Transformation in Healthcare: Innovation, Strategies & Processes Page 18 of 29

systems that are good means for drawing data together around an individual, no
matter which different system it might actually reside in.

Now in that situation, let's say that clinician takes the decision to refer the patient for
some additional tests or some treatment. Some of the treatment might be taking
place within the hospital, some of it they may have to go to a different institution. At
this point, the interoperability tools, some of them have their own workflow engines,
they can trigger off these various tasks, but if that is not available, an independent
robotic process automation (RPA) tool can also be used to trigger this task. The
RPAs can send off emails, generate letters, write information to into other systems
and trigger of any other action that is required from that one single decision. And you
can see how an RPA does through one action triggers of many actions and follows
through on it consistently, can vastly improve productivity and coordination within a
healthcare settings.

The next category of productivity tools to consider is those that improve clinical
throughput. Primary among these are decision support tools. Clinicians often have to
take complex decisions based on a very intense dataset. So, for example, they may
be looking at X-rays or CT scans, they might be looking for small lesions or other
anomalies. In these situations, algorithms can help focus their attention on the things
that matter. And the case of an algorithm, working side by side with the clinician to
take better decisions is one of the great aspects of productivity that's being achieved
through some of these modern technologies.

There are other decision support tools. Sometimes, clinicians have to take traverse a
very complex decision tree to workout a precise diagnosis or a particular course of
treatment. This is particularly the case in disciplines like dermatology, where it can
often be difficult to arrive at a diagnosis or work out a course of treatment. So, these
decision tools and decision trees help clinicians take consistent, high-quality
decisions every time and thus improve the productivity and improve patient care.

Other tools can be used, for example, for patient self-service. So, for example, when
a patient comes to an appointment, often they have to go through a review, they
have to fill an assessment form, they have to complete checklists and not all patients
are able to do this independently. One can use technology to guide patients through
this process and support them without having to take time away from a nurse or
doctor or healthcare assistant to support them. This can be done through technology
like kiosks, digital forms or even smart speakers have been used to support patients
through the self-assessment process.

The third category that we can consider here are surgical tools I briefly mentioned,
suturing robots in a previous video. These speed up the process of post-operative
suturing so that an experienced surgeon doesn't have to spend his or her time
completing the surgical activity and can leave this routine part of the process to a
surgical robot. But the other interesting development in this area, is the Da Vinci
robot, which has now been around for a while. The Da Vinci robot enables surgeons

Digital Transformation in Healthcare: Innovation, Strategies & Processes Page 19 of 29

to perform movements that were not humanly possible, such as the ability to rotate a
wrist through 360 degrees to perform a certain surgical movement. Furthermore, the
precision and the sensitivity of the robot ensures that very delicate surgical
manoeuvres can be performed accurately every time. Provided, of course, that the
surgeon has been suitably trained to use the Da Vinci robot. The final category to
consider here, is digital forms to speed up form filling data capture to ensure that the
data needs to be captured only once and to feed multiple systems. You could also
have digital displays by a patient bedside to capture vital signs, to share information
with the patient and to help the patient interact with the hospital, to ensure that their
needs are best served.

The next category of tools to consider are those, that improve patient safety. One
could argue that patient safety is probably the most important aspect of everything
that the hospital does. And indeed, all of the productivity tools that we've discussed
so far, tend to have some element of increasing patient safety. But often, that is not
their primary objective. You should be no doubt that any tool that's approved for use
does have to go through a clinical safety and a clinical risk audit in order to be
approved.

But the purpose of that order it is not necessarily to show that patient safety is
increased by using this tool, but it is just to demonstrate conclusively that patient
safety is not adversely affected by using some of these productivity tools. So,
productivity tools in themselves might not be sufficient for improving patient safety.
And in fact, a lot of research shows that less than five per cent of productivity tools
that are approved by hospital managers are related to improving the patient
experience or to improving patient safety. We have a separate video coming up,
where we get into the details of the current state of play in patient safety.

There is a company in the UK that has pioneered the electronic documentation and
management of patient safety incidents. This company is called Datix and their name
has become synonymous with patient safety incident recording. There are now
predictive tools that can also highlight to clinicians, particular safety hotspots that
they need to pay attention to, in order to reduce patient safety risks. This is an
important topic and we will spend some more time getting into the details of how
patient safety can be enhanced by the use of technology.

The next category of productivity tools are those that improve the utilisation of high-
value assets. Hospitals invest a lot of money in expensive equipment, some of this
equipment might be large and immovable. But there is an equally or greater number
of devices that are expensive and can be moved around. These are portable
scanners, investigative instruments, scopes, trolleys and even special purpose
wheelchairs or gurneys. Clinicians and other hospital workers can waste a lot of time
looking for equipment when they need it. And in order to prevent this wastage of
time, hospitals tend to over-invest and buy more of this equipment then they might
normally need.

Digital Transformation in Healthcare: Innovation, Strategies & Processes Page 20 of 29

To avoid that, more and more hospitals are using radio frequency identification tags
or RFID tags for short. These are active devices that can be attached to every asset
and they can provide information to the hospital wireless network to, say, give the
location and to state whether, they are in use or not. This becomes a good way to
identify what is the nearest available device and whether it can be brought into use
for a particular situation. So, if clinician needs a scanner or a wheelchair, they know
where the nearest one is available and they can send someone to fetch it. The next
set of productivity tools relate to the utilisation of people.

Hospitals are open all the time 24 hours of the day, seven days a week, 365 days of
the year. And when they are opened, they must at all times be start with enough
people to deliver safe care. These demand patterns vary by the time of day and the
day of the week, they vary by season. So, working out the optimal staffing mix at all
times is a real critical success factor for hospital managers to ensure that the
hospital performs at the optimal level and they manage all of the costs of the human
resource that is required for staffing. It is really valuable to have a clever e-Rostering
system as they're called, to determine those staffing patterns. These systems do
simulations of what is going to be the likely demand for the next two or three days,
based on all of the data that they had of the past, it can then look at the database of
skills of people who are available by skill mix and take into factor holidays, absences,
sick leave, etc. to identify what is the pattern of a fully employed staff that's
available? What staff need to be brought in from the flexible resource pool or the
staffing agencies? And identify any critical gaps and highlight them to management
for a decision.

The final category is eliminating waste. Undoubtedly, many of the productivity tools
that we've discussed so far, help eliminate waste in some form or the other. But
there are other technologies, that are available for specifically aimed at eliminating
waste in a hospital setting. Energy management is a particularly important issue and
hospitals are huge users of energy. Doing energy audits using technologies for heat
mapping can identify areas that the estates' team needs to address to ensure that
there is no unnecessary energy loss and that hospital buildings are well insulated,
provide the right level of heating light and the right ambience for the proper care of
patients.

There are even other simple technologies like setting up trading or barter portals for
hospitals to exchange furniture equipment that they don't need. In the past, this is to
go into landfill, but by putting unused equipment of furniture on the website, other
hospitals can bid for it or even take that equipment for free. And this helps reduce a
lot of unnecessary waste as well. So, as you can appreciate, hospital managers
have a huge portfolio choices and they must make a selection. Which of these
productivity tools can they use? What fits within their budget? What might have the
maximum impact in the setting? They need to do that assessment and build a
business case for adoption. There are tools like simulation which help hospital
managers take these decisions. They can work out what is the potential impact of

Digital Transformation in Healthcare: Innovation, Strategies & Processes Page 21 of 29

any process change or technological change. And do a multi-factor analysis such as,
how would safety be improved? How would costs be reduced and so on and so forth.

In some countries, like Singapore, the Ministry of Health has invested in building
digital twins of some of their key hospitals. This enables them to conduct simulations
that scale of various transformation projects that they may have in mind. The digital
twin helps them make a quick impact assessment or which interventions or which
transformation strategies are likely to benefit the citizens of Singapore optimally.

Video 9: Predictive and prescriptive analytics [15:11]

The availability of huge amounts of data and cloud computing is enabling a new
class of productivity tools. Predictive and prescriptive analytics work on the same
sorts of principles. Given a sufficient amount of data and a learning mechanism or a
learning model, a computer can then generate alternative future scenarios and the
probabilities of the scenarios coming into fruition. This is called predictive analytics. If
you run a predictive analytics model for long enough and start collecting data about
how its predictions actually panned out in reality, it starts developing greater and
greater confidence about a likely future scenario, given a set of input data. As this
confidence builds up, it can start making very definitive recommendations on what
needs to be done next in a particular situation, that is, it offers a definite decision or
an action. This is prescriptive analytics.

I would like to offer you some use cases for predictive and prescriptive analytics in
healthcare. Just bear in mind that this is not a comprehensive list. As time goes on, I
expect other interesting options will emerge and could be added to this list. The first
of these options is the ability to predict or prevent an admission to a hospital. The
second is the ability to prevent a readmission after 30, 60 or 90 days. You then have
tools that are helping coordinate patient care through technologies like next best
action (NBA).

You have the option to predict the demand at a hospital or at an emergency

department on a particular day of the week or time of the year. You have predictive
tools that can identify people who may be at risk of self-harm or suicide. And in a
less intense use case, you could predict people who, maybe, are likely to miss a
hospital appointment, what is called did not attend (DNA). And finally, you have
population health management (PHM) tools that can help predict what is the future
requirement for resources within a local health system to help improve the overall
health of that population.

Let's first look at, the use case for preventing, or perhaps even expediting an
admission to a hospital or emergency department. There has been a lot of work in
this area on heart failure. There is a clinical protocol that shows that patients who
have had an overnight weight gain, a certain overnight weight gain could be at a risk
of hospital admission for heart failure. People with these diagnoses have to monitor

Digital Transformation in Healthcare: Innovation, Strategies & Processes Page 22 of 29

their weight every day. A sudden overnight weight change means their heart is
beginning to decompensate and this is usually signified by increased fluid retention
in the body. This has picked up and when this weight gain is detected, the patients
are asked to take a diuretic pill to eliminate that excess fluid from their body and
stabilise their health and avoid that emergency admission. This protocol works really
well and you can see it's a prescriptive analytics. If you've had this overnight weight
gain, if you're showing symptoms of fluid retention, take the pill.

Another company has found a novel way to do the same prediction. They do it
through voice analysis. What they discovered is, if a patient speaks into their app
and says, ‘Good morning, I'm feeling really well today’ or ‘I'm not feeling so well
today’, they can figure out from the quality of the voice, how much fluid retention
there is in the body because they found a very close correlation between the
changes in the voice and the amount of fluid that is accumulating. And their
algorithm has the same predictive power for this risk as taking weight on a daily
basis.

There have been other prediction models used for many years in healthcare. The
most famous of these is the Framingham cardiovascular risk model. It is based on
about 70 years of data collected from people in the town of Framingham in the US,
calculates the risk factors that would predict a potential cardiovascular event in ten
years time. The Framingham Risk Score is used to take clinical decisions on
treatment of people with a potential heart condition.

The second use case is preventing readmissions in 30, 60 or 90 days. Readmissions

are a particularly bothersome problem for hospitals and also for payers. Some time
ago, hospitals were not so concerned about readmissions and you would find quite
often that a patient was discharged, needed to be readmitted and treated again for
the same situation. Payers started getting bothered about this because it seemed
that hospitals were double or triple treating and then billing for these patients. And
therefore, started imposing fines or penalties for patients who are being readmitted
with the same condition. This was then encouraging hospitals to focus on
discharging patients properly and safely at the right time. that has helped, it doesn't
always prevent many of the readmissions that still continue to take place. So, there
has been a great effort placed on trying to predict, what sort of people are likely to be
readmitted in 30, 60 or 90 days.

In the UK, there was a model developed many years ago called the patients at risk of
readmission (PARR), which was trying to identify these risks from the hospital data.
They found that the risk model was not strong enough in its predictive power. So,
they combined the hospital data with primary care data and created PARR+. It was
hoped that that would improve its predictive power but it didn't do that sufficiently.
So, they combined the hospital data with the primary care data with social care data
and created PARR++. And even that was not predicting the readmission risks
sufficiently. So, there is a cautionary tale in this, whilst predictive models might work

Digital Transformation in Healthcare: Innovation, Strategies & Processes Page 23 of 29

at a population level, they could still be frightfully inaccurate or useful at an individual
level. So, the goal for trying to find this prediction model still continues.

The third use case is about coordinating patient care, particularly for people who
have complex needs. The tools used for coordination include, called next best action
or NBA and we've mentioned this before. NBA tools have originated in software like
customer relationship management and perhaps other kinds of software as well. And
what it basically does is it pulls up data about the current situation of a particular
subject that might be a customer or a patient, it looks at all of the historical data on
decisions that were taken in such a situation for similar people. And it also will
consider the outcomes of each other decisions and combine all of that information to
create a ranked list of, what are the options for the next best action (NBA), in this
situation, for this person. And it may offer up a list of three, four, five potential actions
for the person to take. So, if a healthcare professional is looking at, say, a list of
three actions, they might pick the second one and the NBA system will help track the
execution and what were the outcomes from that decision. So, next best action
(NBA) is, in some senses, it's a part predictive part prescriptive tool and that can
have great power in coordinating care for people.

The fourth use case is in predicting hospital demand or an emergency department

demand. When hospitals start collecting large data sets, it's then possible to start
creating some predictions about the number of people and the types of demand that
will arrive at the emergency department's door. Some of these things are fairly
common sense if you work in a hospital. You know that there are more people on a
Friday evening and on weekends, you know that, for example, at festival times, there
are more people who come into hospital due to overindulgence in food, drink or other
behaviours that can cause harm or risk. Wet weather or extremely cold weather can
also bring people with a variety of illnesses or from traffic accidents. What is slightly
less well known is sometimes, in countries, where the clocks change and the clocks
go back and people have less sleep. That causes more accidents and more health
events the next day. As you start trolling through the data and start drawing
connections between the external events and arrivals at the hospital, predictive tools
can give other early indicators of the volume and type of demand that might come to
a hospital.

The fifth use case is around preventing harm, self-harm or suicide in hospitals. The
whole development of smart sensors in the Internet of Things (IoT), low power radios
and creating mesh networks they are bringing huge benefits to managing the safety
of patients in a hospital or a home. So, sensors like bed sensors, fall detectors,
movement sensors or passive sensors to monitor vital signs like heart rate,
respiration rate, level of agitation, body temperature. These are helping the
monitoring of patients in a non-invasive way to keep them safe. Philips Healthcare
has a vast portfolio of such sensors, that are used both in a hospital or home setting.
And then you have companies like OxyHealth, which are using computer vision and
image analysis to help keep patients safe in an acute mental health setting. They are
able to identify through the algorithms, patients who might have harmed themselves

Digital Transformation in Healthcare: Innovation, Strategies & Processes Page 24 of 29

or might be at risk of harm and draw the attention of staff to people who might need
and they do so by observing the patients in a non-invasive manner.

The sixth use case is, around reducing missed appointments or patients who did not
attend an appointment, it's called DNA. DNAs are a big cause of waste in healthcare.
If a patient misses an appointment and that capacity is then lost forever, that you
know, that time has gone, it is never available again. In order to try and reduce this
waste, hospitals will try and overbook a clinic because they expect their five to ten
per cent of patients will not show up. It's a bit like, what the airlines do. This is to help
ensure that the clinic is fully utilised.

And so, what happens very often is, everybody shows up and the clinic is extremely
busy and it leads to exhaustion of the clinical staff. So, there is a delicate balance.
There is a lot of follow up that takes place to ensure that patients attend and to try
and bring down these DNA rates. So, there are a lot of follow up calls, text
messages, reminders to patients that they have an appointment. In order to try and
reduce that follow up effort, researchers at University College London (UCL), have
come up with a predictable algorithm that can reduce that follow-up effort by
targeting specific patients. That's reduced the effort by about 50 per cent. But clearly,
there is a lot more that needs to be done even to solve this simple problem.

The seventh use case is Population Health Management or PHM. Population health
management (PHM) uses vast quantities of historical data to try and understand
what are the current demands in a healthcare system and how those health
demands might change in the future. Local authorities and service planners use
PHM systems to try and understand what are the various variations in healthcare
due to where people live, their demographic situation and they use those insights to
plan new services to ensure that the population of that whole area, the health is
managed really well. So, you can consider PHM systems as a form of predictive
analytics, for population-level strategic resource management. They can be used in
various tactical situations as well. So, for example, during the COVID-19 pandemic,
PHM systems were used to try and identify areas where there were larger
populations of people who might be more vulnerable to an infection and to identify
situations where there needed to be a specific campaign to invite more people to
come forward for vaccinations.

As I said earlier, these use cases are not exhaustive. In a separate video, we will be
looking at patient safety and the role that predictive and prescriptive analytics have
to play in this very important area.

Video 10: Patient safety [12:48]

The primary goal of our hospital is to provide the right package of care for a patient.
Often, when a patient is admitted to hospital, it is a high-risk situation. It could be life
or death. In these high-pressure environments, the doctors and nurses have to take

Digital Transformation in Healthcare: Innovation, Strategies & Processes Page 25 of 29

decisions on what do and how to treat a patient and the goal is to get the patient
better. But often patients also encounter unnecessary harm or risk at a hospital. And
so patient safety looks at the other side of the experience of being in hospital. It's the
other side of the coin. If a Boeing 747 full of passengers were to crash every single
day in the USA, would that be news?

There is a lot of preventable harm that takes place in hospitals. In the US, that
numbers about 160,000 preventable cases every single year. It used to be more
than that, but it's come down about 20 per cent in five years, so clearly there's a lot
of good work that's being done. But this problem is everywhere you look even in
England, there are about 750 preventable deaths in hospitals in England, every
single month. So, in this final video on hospital transformation, we would like to focus
on patient safety rather than just productivity techniques alone.

Patient safety ought to be an integral part of any transformation programme and this
is because our focus on patient safety does actually have productivity benefits. By
running a safe hospital, you reduce the number of infections, you reduce the number
of rework, readmissions go down, the length of stay in hospital is reduced. So, a
focus on patient safety does have massive productivity benefits besides, the obvious
benefits for the patient as well. So, if you are currently in a role or likely to be in a
future role around a hospital transformation. I would ask you to take a very strong
patient safety and patient experience perspective because that would really help
achieve the long-term transformational goals of your programme. Patient safety
systems have been around for a while. It's not that hospitals did not try to offer a safe
experience and these are often consisted of pen and paper, there have been
surgical safety checklist for a very long-time, to just ensure that surgeons and the
whole surgical staff had been properly prepared for each procedure.

About 20 years ago, there was a lawyer who did a lot of work for hospitals in the
National Health Service (NHS) in England, defending them from claims that were
arising national patient safety incidents. He started documenting the background of
all of these claims using a computer system. The basic premise of the system he
developed was, if an event takes place, let's document all of the cases carefully, let's
document all aspects of that incident while everyone's memory about it is still fresh.
So, if subsequently a claim is brought against the hospital, you have all of the
documentation in place to either defend against the claim or indeed, if a mistake has
been made by the hospital to admit liability and settle the claim as quickly as
possible without getting involved in a lengthy legal process. So, this system once the
documentation starts getting built up, every case that's documented gives the ability
to learn from this situation and to avoid that mistake from happening again.

And this is how the system known as Datix was born. Datix today, provides a lot
more than a tool to simply record incidents or to handle claims. It has analytical tools
that helps hospital managers and teams understand what might be incident hotspots
and whether these are related to specific procedures or equipment or locations or
even teams. It is able to highlight these hotspots so that action can be focussed on

Digital Transformation in Healthcare: Innovation, Strategies & Processes Page 26 of 29

resolving the underlying root causes of those errors. And it stimulates learning
organisation that is constantly driving improvement based on the evidence that
collects about its own errors and improvements. Datix has been very successful.
Over 80 per cent of hospitals in the National Health Service (NHS) now use Datix. It
has also been adopted by the hospitals of the US Department of Defence and the
US Veterans Administration. The true indicator of success is that Datix has now
become a verb.

So, when a safety event takes place, clinicians say, let's Datix it. In the shadow of
Datix's dominance, there were other interesting applications and systems that were
beginning to grow and gain into prominence. These were focussed on ensuring that
the right clinical protocols and pathways were being consistently followed by staff in
hospitals and they did this by a combination of collecting bedside observations
electronically and generating early warning scores or modified early warning scores
called MEWS, through this electronic process.

VitalPAC was one of the first such systems and in the early days an iPod touch and
then an iPhone for nurses to collect bedside observation data on things like vital
science. It also gave them prompts to follow clinical protocols, such as went to
change a catheter or how to clean a particular dressing, it reminded them when
patients need to be reviewed and when their oxygen needed to be checked, it
prompted all of the clinical staff who were involved in the patient's care to do the right
things at the right time. It started collecting all of that data and then they start looking
for patterns, patterns of improvement or patterns of deterioration. If there were
patterns of deterioration, better patient needed help, it would raise an alert to specific
members of staff who are on duty at the time and if that task was not attended to, it
would then get escalated to a superior or to a specialist to come and give attention to
that patient.

The beneficial effect of systems like VitalPAC was soon obvious to everyone and the
results were published in the medical journal Lancet and the British Medical Journal
(BMJ). In one of the cases, the articles documented the impact on norovirus
outbreaks in hospitals. Now, norovirus is commonly known as the winter vomiting
bug. Infections in hospitals spread really fast and it's been a particularly troublesome
problem in winters, when this infection would cause an entire ward, or sometimes all
hospital to be closed down because all the patients got infected. What was
interesting is that hospitals that were using VitalPAC to track patients under
deterioration was able to identify patients with an infection really quickly and then
you would have infection control teams come in to isolate that patient, treat them and
ensure that the infection did not spread. So much so that hospitals that were using
VitalPAC almost never had to close down a ward because of a norovirus infection.

This type of close and systematic observation and management of patients in

hospitals meant that in one London hospital, emergency calls for cardiac arrest of
patients who were within the hospital went down by as much as 70 per cent. This
meant that the hospital experienced a 120 fewer cardiac arrest amongst patients

Digital Transformation in Healthcare: Innovation, Strategies & Processes Page 27 of 29

who were in the hospital. This meant that patients were getting better care and any
problems were being detected really early and were being resolved before they
resulted in a cardiac arrest. The acute medical unit of that hospital had a period of
100 days where there was not a single cardiac arrest in that hospital, in that unit.

The British Medical Journal's (BMJ) Quality and Safety Journal published evidence
that nurses who are being used in these electronic observations and MEWS systems
had reduced all-cause mortality in hospitals by as much as 15 per cent. So, I was
talking to the founder of VitalPAC and he was saying smilingly to me that VitalPAC
might have been the best thing since penicillin. If you visited any conferences lately
or even webinars, you will find that they have been buzzing for quite a while about
big data, machine learning, artificial intelligence (AI). And healthcare conferences
have always been getting quite excited in recent years about biosensors, quantified
self and all of these modern and amazing technologies.

But there's a difference now, that buzz has changed into some sort of action, with
record levels of investment from both from venture capital and big tech. And there
has been renewed interest from the giants like Microsoft, Google and even Amazon,
very recently pouring millions, even billions of dollars into healthcare and health tech
related investments. So, this is a good moment to come back to that problem of
sepsis, we've seen so far, what impact could be had by taking a low-tech approach
with the flow coaching model to resolve the problem of sepsis management. You've
also had the opportunity to consider how building a discrete event simulation (DES)
model could help in the improvement of sepsis management.

So, imagine what could happen if you threw the power of big tech and all of these
exciting new technologies on the management of sepsis. Well, hopefully you don't
have to wait too long. The Institute of Global Health Innovation (IGHI) at Imperial
College is doing exactly that with Google DeepMind. They're taking data from
electronic medical records (EMR), the deep Learning network and the app
associated with that is generating alerts of people who might be at risk. And then it's
directing action, based on the alerts that are being raised. Now, bear in mind this is
still a research project. There is a lot more work to be done. It started in 2016 and
there is ongoing work to keep refining the algorithms and let's see where this
research gets us to.

Video 11: Module wrap-up [04:19]

As we wrap up this module, let's reflect on the key learnings from it. The hospital of
the future needs to be quite different from the hospital of today, which is very deeply
rooted in the hospital of the past. But our healthcare needs have changed
substantially in the past few years and what we need from our hospital systems is
very different. The scale of change that's required in hospitals is not incremental, it is
transformational. And there is no doubt that technology is required to achieve this
transformational change. That said, the most hospitals are still focusing their

Digital Transformation in Healthcare: Innovation, Strategies & Processes Page 28 of 29

transformation efforts on getting the most modern technology, improving efficiency
and productivity and even improving profitability. It could take an earthquake to
rethink the role of the future hospital.

Hospital leaders face daily challenges in trying to achieve the quadruple bottom line
of better patient experience, achieving better patient outcomes through quality and
safety programmes, providing a better experience for their clinicians and achieving
financial balance by managing their costs. Some the leading hospital troupes have
instituted innovation management capabilities within their organisations and the role
of these functions is to create the capability and the resource to systematically drive
new innovations and transformations through their organisations.

We covered different tools and techniques that hospitals have to support their
journey of transformation. We looked at flow coaching, an analogue model of
understanding and developing creative solutions to problems in day-to-day
operations and in some deep-rooted issues that have often been difficult to
diagnose.

We then looked at discrete event simulation (DES) to try and assess the impact of
solutions that were coming out of flow coaching, to understand the case for
implementation and to make some choices based on hard data on which options
should be implemented. And we've then looked at predictive and prescriptive
analytics and some of the opportunities that group of solutions offers for
transformations of various aspects of a hospital's services.

We then started bringing all of these together in the context of sepsis management
and focusing for a moment on the sepsis challenge, we saw concrete evidence of
how flow coaching can help improve services for detecting and managing sepsis
hospitals. We used a simulation model, to understand the key levers that can be
controlled to improve sepsis management. And we then saw, how tools like VitalPAC
that allow electronic observation recording and early warning scores to identify
people who may be experiencing sepsis in a hospital. and those tools will then drive
the alerts and the interventions that are required to help keep those patients safe.

And finally, we took a glimpse at what might be the future of sepsis management of
using AI and detailed EMR data to predict patients who may be at risk of sepsis. But
what we also saw in this case is that research has been going on for a long-time and
it's not quite finished yet. So, it's worth asking the question. Do we really need
something like AI to solve sepsis management?

-----------------------------------------------------END-----------------------------------------------------

Digital Transformation in Healthcare: Innovation, Strategies & Processes Page 29 of 29