This article is based on a Webinar titled “How to build knowledge based collaboration at scale in healthcare” which you can WATCH HERE. You can also access a white paper on the subject HERE.
- Most of my data is on in-house servers which I’m reluctant to fiddle with, so I was wondering at what scale of data manipulation and sharing I should think about some sort of cloud set up?
Cloud can bring benefits and efficiencies to any level of data storage, sharing and analytics set up, big or small. Ask yourself some of these questions to understand how ready you are to move to cloud or upscale your set up:
- At what point are you in the amortisation of costs for your ‘on premise’ IT set up? The nearer you are to the end of your amortisation cycle, the more easy the decision can be to go cloud as you won’t be paying for two set ups simultaneously or wasting what is left on your payments for your in house set up. Notwithstanding this, often the utility of cloud will outweigh in house set ups even if there is a lot of time left in your finance amortisation cycle.
- What level of data sharing are you attempting with other organisations and how are you doing that – usually it means you will be using the cloud in some way anyway. If sharing and collaboration on analysis is a key part of your working paradigm cloud is going to help a lot.
- When you are sharing data are you finding issues with what cloud vendor your collaborators are using compared to what you use? Secure cloud exchanges are starting to solve this problem for a lot of organisations.
- If you have a big investment in an ’on premise’ set up, are your financial managers aware of what a shift to cloud means for the financial management of your organisation, i.e. You shift cost and cashflow a lot as your infrastructure for data storage and management shifts from a 3-5-year amortisation cycle with most of the cash spent up front (Capex), to a pay as you go operational cycle (Opex). Does that help or hinder your organisation at this point of time?
- What are your partners and competitors doing? Is it important to keep up?
Remember that cloud for nearly all organisations today is not an either/or choice. Most organisations start with a hybrid cloud strategy and work their way up.
- Can I start slow and do bits of both and grow out?
Yes, it’s how most organisations make the journey and there is a spectrum upon which you operate. Indeed, it may be that you always retain some form of on premise set up, which talks to your cloud set up, and that will be the most efficient and secure for what your organisation needs. Nearly nine out of ten organisations today expect to have a significant or measurable on-premise environment in three years.
Naturally, most organisations have an on-premise-first mindset when they start formulating their hybrid cloud strategies. Most cloud service providers offer solutions that integrate easily with an organisations on-premise environment.
When you start you should plan for where you are heading though. Set out your objectives and a plan for how to get there. In very broad terms you can mix the following for the most effective set up, or migration to more cloud-based services:
- On Premise: Servers are on premise. You may need to keep some of your set up this way for ever for speed of some data analysis and retrieval, or for organisational or government data governance reasons
- Hybrid: Mix of on and off premise. The off premise can be ‘near premise’ for the lowest latency (highest speed of down and upload), and data governance issues that restrict data to being physically in certain locations. Each has different latency and cost.
- Cloud: is off premise entirely, but can be nearby for latency and governance
The most difficult aspect of adopting cloud services is usually application migrations. Just over thirty percent of organisations report that the preparation work—including building the business case—for application migration is hard, while another thirty percent say the the process for redesigning and actually moving these applications to cloud-based platforms is the greatest challenge. There is no one-size-fits-all approach so partnering with the right organisation to help you migrate is important.
- Can you give me an example of an organisation and set up that is getting a lot out of cloud for data management and collaboration and what they did?
A good local example of a healthcare organisation using cloud to manage their data and collaboration is the Children’s Cancer Institute , an independent research centre wholly devoted to ending childhood cancers. The Institute runs a program called ‘Zero Childhood Cancer’ which optimises cloud in order to share data across multiple organisations. This is essential in order to share and use highly personalised genomic data to cure high risk childhood cancers.
Twenty percent of children diagnosed with cancer end up with no or very difficult treatment options.
The ‘Zero Childhood Cancer’ program is using the latest in genome technology, like whole genome sequencing and RNA sequencing to understand the genetic makeup of every patient’s tumour, with a view to finding bespoke treatments. To get this done the Institute will generate about one terabyte of data for each patient and in order to find what might work for each patient they need to search databases across about 30 other institutions around the world.
On a per patient basis the cloud is ideal because giant amounts of data is analysed in bursts, and lots of data needs to be shared quickly with other databases held by other institutions. In the case of the the Institute, they use nearby cloud, hybrid cloud and an emerging technique of multi-cloud, whereby they are able to work reasonably seamlessly between different cloud vendors. The latter ability is proving hugely useful as nearly all institutions run on different clouds, with different vendors such as Amazon, Microsoft, Google, and up to 200 others.
“We can compare our rare cancer patient data to other rare patients, whether it’s in Philadelphia, Memphis, Heidelberg, or wherever that patient may have may have been diagnosed and treated” says Associate Professor Mark Cowley.
Using this technique, the Institute has been able to identify drugs already in use which have proven effective in very rare local cases of childhood cancer.
“This ability is the changing landscape of this data. No longer are we just looking inside just our institutions. We’re looking to all the partners around the world as to how we can share data in near real time so we can make improved recommendations for treatment for these kids.”
Another innovative aspect of the set up at the Institute is that they can now run their own analytics tools wherever data resides, whether that be in one vendor cloud or another, or in one city or country or another. One way of achieving this was using a ‘secure cloud exchange’ – in this case Equinix – which is home tomore than 2,900 cloud and IT vendors and has locations all over the world (more than 210 data centres in 26 countries serving 56 markets) for providing nearby facilities which can switch directly between cloud providers and other cloud based services. This makes speed and connecting easier and less expensive.
- OK, exchanging data technically feels like it’s getting a lot easier, but aren’t issues of privacy, governance and regulation going to hold all this back anyway?
Regulations are generally always behind when it comes to new technology and to some extent there is no question that data sharing in the cloud is one of those areas. Different countries also have very different regulations. Some are much more behind, some are more restrictive.
But it works in a spectrum. In the case of Children’s Cancer Institute, the group has identified the key data partners it needs to share data with and has worked out the protocols of sharing in spite of geo differences in governance and regulation.
In other organisations, for example a multinational pharmaceutical company doing research, the ability to use data which has a multi country source and share it with outside organisations can pose a challenge, because some countries data cannot be easily shared with another country or another organisation.
- How do cloud vendors deal with issues around health data sovereignty?
There are very good reasons why as a hospital or as a state, you don’t want your patient records necessarily leaving the state or country. You will want to avail yourself of the computing resources or analytics tools on those records. But you absolutely want to make sure that they’re staying within certain boundaries of geography and governance.
Managing Director of Equinix Australia Guy Danskine, says that globally Equinix experienced this issue very clearly in Europe a few years ago when General Data Protection Regulation (GDPR) was introduced.
“We had a number of healthcare companies looking for ways to restrict the flow of data so that they could utilise and leverage the tools they needed while ensuring that it was only in the locations where they needed to be,” he said.
Danskine says that in Australia state sovereignty has been important, but in certain markets they’ve been able to design a hybrid cloud setup, which means that certain data remains within a local council area, while all the tools are made available to process that data.
- How do organisations ensure that the data they are sharing is going to be used ethically?
When pharmaceutical companies started doing clinical trial research 50 years ago there was no central repository of the clinical trials that organisations could access. Data was highly siloed and therefore far less accessible and useful. Today, however, there are public databases managed by multi-country collaborative groups, with appropriate governance and controls built in and constantly updated.
Dr Bernard Ng, Head of Global Medical & Clinical Affairs at Bayer Consumer Health, thinks that something similar will happen in future with a lot of medical data, not only in terms of private data, but also public data.
“In today’s technology and today’s world, we, even as personal consumers, are sharing a lot of our data publicly,” he told Wild Health. “I want to believe that people are using it for the good of the society for the advancement of science. It will always be a two edged sword where people can always abuse it. But if we let that stop us from sharing the data, we will not be able to achieve the potential good that may come out of sharing the data.”
- What are the cost implications of moving to cloud?
Generally, the simple movement of data storage and applications offsite will end up costing an organisation less than buying, maintaining and installing on site servers. But it’s in how you deploy outwards with the cloud that you really start to get return on your effort and investment. For starters, you can scale up and down to your exact needs depending on what applications you are running and when, so you only pay for what you use. And you never hit a limit, as you can with on premise installations. Guy Danskine, Managing Director of Equinix Australia, which is a global interconnection and data centre provider, says that they have an expression for this: “own the base and rent the spike”.
Then you can move into sophisticated means of sharing, using near premise when you need better latency, all the way to something sophisticated, like the multicloud applications and connections that Children’s Cancer Institute uses.
Cloud can end up costing you net more than you used to spend on your in house set up, but nearly always the return on your investment is a high multiple of what you once had in terms of data manipulation and research progress. This is why people will often refer to use of the cloud as “transformational” for some organisations. The agility, speed and sharing capability can transform a groups capability within months.
Telstra Enterprise’s Principal Cloud Strategist, Kent Ramchand, puts it another way. He says it’s not so much taking out cost as adding enormous value to your business by increasing your ability to collaborate and share data.
[“Cloud] is not just cost takeout. It’s really adding value as part of a digital ecosystem,” he said.
Ramchand points to a collection of Telstra Health companies which have been growing by acquisition and using cloud in order to share data quickly between acquired companies to gain momentum.
“ Containerised workloads that require massive scale very quickly, are absolutely perfect for cloud computing,” says Danskine. “It’s really about identifying the workloads and what kind of characteristics they require. And then that determines the best place for applying your cloud services.”
- What about AI health applications, data processing and cloud?
Dr Denis Bauer, Head Cloud Computing, Bioinformatics at CSIRO, says her organisation has a “mandate to be the collaboration hub [of AI ] in Australia, bringing together universities, small businesses and industry at large”, and that cloud is a key part of meeting that mandate these days.
She says that AI and machine learning algorithms that have been around for years flourish in a cloud environment as a result of the sheer amount of computing power the cloud can provide, and its agility in being able to scale up from research systems to full production systems, and going from small experimental data sets to full scale, population data sets.
Dr Bauer works with a machine learning library called VariantSpark which can analyse population scale genomic data.
“We recently published a paper that was the world’s first to demonstrate that we can apply machine learning on one trillion data points, which is quite staggering, and definitely where the future is when we think of population scale genome data analysis,” she said.
Dr Bauer is working on the idea that you can have one algorithm that the community continues to improve, but it’s deployed on all the different vendor cloud platforms. This can be done using a ‘secure cloud exchange’ service like Children’s Cancer Institute does.
- Should you be architecting specifically for the utility of the cloud to optimise what you are doing?
If you’re getting started there is nothing wrong with ‘lift and shift’, which is essentially just porting some data and applications onto the cloud, according to Dr Bauer. That can achieve a certain level of agility and efficiency in data management above and beyond simply an on premise set up.
But, says Dr Bauer, once you are familiar with your cloud ecosystem and ready, a lot more can be achieved if you design your set up specifically with cloud functionality in mind.
“Native cloud applications offer so much more in terms of scalability and modularity of data management and in terms of bringing in different components from different vendors into your system,” she told Wild Health.
“Building a new system that only can exist in the cloud is really where a lot of big benefits are,” she said.
- From the experts: In one sentence (perhaps two) why should I consider cloud more carefully for my healthcare organisation?
“Cloud is not a technology problem. It is a fundamentally different mindset of delivering and doing research. It’s more how well do we share the data, how easily do we make the data available and how do well we analyse the data?” – Dr Denis Bauer, Head Cloud Computing, Bioinformatics at CSIRO
“You need to keep your options open so a hybrid cloud pattern of, you know, having private cloud and public cloud for those spikes (that were mentioned in this article) are really the key to making sure that you have all the options available to you no matter what application is flavour of the month at any one point in time.” – Kent Ramchand, PhD Candidate & Principal Cloud Strategist at Telstra Enterprise
“We used to write snail mail, and now we all use email. Is snail mail still around? Yes, but everyone uses email. There’s no going back. That’s how I see cloud moving forward” – Dr Bernard Ng, Head of Global Medical & Clinical Affairs at Bayer Consumer Health
“From a very practical perspective we first shifted to cloud because I needed a single core server for 12 hours, a 32 core server for 37 minutes and so on. It’s that sort of flexibility in our workflows to be able to do tens of thousands of hours of computing very, very rapidly and bring data together in a way that you could never get in your in our HPC environment” – Mark Cowley, Computational Biologist at Children’s Cancer Institute
“Equinix worked with a healthcare provider in the US called Sentara because they required connectivity to AWS through us. But it wasn’t for a lot of the reasons you’ve read about above. It was so that their patients could interact with their hospital rooms through the Alexa, personal assistant. When you think about innovation, health and technology, cloud is the way to bring it all together no matter the problem” – Guy Danskine, Managing Director at Equinix Australia
This article is based on a Webinar titled “How to build knowledge based collaboration at scale in healthcare” which you can WATCH HERE. You can also access a white paper on the subject HERE.
