The ambulance arrives at the scene of an accident. A critically injured person lies bleeding in the road. EMT personnel must make treatment decisions, and fast. But all they know at this point is the patient’s identification. Is she allergic to any drugs? Does she have a history of heart-attack or stroke? What other medications is she currently taking?

In an emergency situation, access to a patient’s medical history could change the treatment plan — and avoid fatal mistakes.

“In the United Kingdom, about 22,000 people a year die because data isn’t always available when it’s required,” said Richard Jarvis (pictured), chief analytics officer of EMIS Health (EMIS Group PLC).

Jarvis spoke with Dave Vellante (@dvellante), host of theCUBE, SiliconANGLE Media’s mobile livestreaming studio, during theCUBE’s special coverage of the AWS public sector portfolio at the company’s London headquarters. They discussed how EMIS is applying new technology to the healthcare market (see the full interview with transcript here). (* Disclosure below.)

[Editor’s note: The following answers have been condensed for clarity.]

Let’s talk about EMIS Health. What is the company all about?

Jarvis: EMIS Health is a software company. It specializes in providing healthcare solutions for a range of clinical settings. Our biggest customers are primary care; we cover about 60% of the U.K. population. This means we have many different care settings, and our big challenge is to bring those together and make the most of that healthcare data.

One of the frustrating things for patients is not having access to their personal records. Can you help solve that problem? 

Jarvis: Absolutely. Access is a problem that exists world over. But, the most important thing to get right first is the security. What your doctor can see might be sensitive, and you might not want to share that with your pharmacist, for example. So, understanding the security model for all the data is the number one concern, but access is also important.

If we can bring the data together so, for example, your allergies are available to an ambulance service when they need to administer a drug, then that will help improve people’s care and save lives. We can do that, and we can do that because we are using the power of the cloud to bring that data together.

The National Health Service of England is the largest single-payer healthcare system in the world, and it has just launched has a major modernization initiative for digital transformation. You are a player in that ecosystem. What’s happening? 

Jarvis: Healthcare needs to undergo the same revolution as other industries, and at EMIS we’re leading that initiative. We’ve spent a lot of time looking at how our customers use our software; how we can use data and cloud services to improve the quality of what they get; and also, the associated cost of production.

What we’re looking to do is to take a lot of the legacy technology — that is very powerful and very heavily used, but also critical to patient care in the U.K. — and moving that into a cloud-based solution.

Doctors know what digital technologies look like when they are good, and they want to have that same experience with their work. Access to data, easy to log onto, secure …  we’re doing that. And not just EMIS, but the whole healthcare economy is moving in that direction.

I’m going to put forth a premise: I would say this transformation would be impossible without the cloud. Is that true, and why? 

Jarvis: Yes, I think it’s impossible without the cloud. One, because of cost and security. But also, the scalability. You need to start small — maybe a few dozen servers — and then your adoption goes up. The cloud gives you that scalability, whereas running your own data center is a distraction from providing patient services, which is really our focus.

Can you talk about security and privacy in healthcare?

Jarvis: Security and privacy are very different, but they are two important parts of the same puzzle. With cloud it is straightforward to prove to a very high level that security is good. There are some serious government organizations who have trusted the cloud with their data, which gives you a degree of confidence that you can trust it too, and there’s some great publications from people like the NHS on what you should do to make sure the cloud is secure.

Privacy is different. We hold about half a trillion patient records. How do we ensure that only the right person can see the right record, at the right time? Currently, researchers have to extract the data for the patients they’re allowed to access and move it into a silo. Then they do their research on that silo. The problem is that if they need additional data, they must do the process again. It is very slow.

To solve this, we’re trying to leave the data in the cloud but apply very rigorous privacy and security constraints on top of it to audit who’s using the data and when. Because, for example, if you are researching cancer, that does not mean you get access to diabetes information.

There are some fantastic tools in AWS that allow you to do that, and then there’s also the ability to extend AWS with your own tools to enhance that further; through APIs, through open-source software. By programming your own tools, you can use all the best methods for your problem.

Can you be more specific about the tools EMIS uses?

Jarvis: Our architecture is based on bringing all the data into S3, which is Amazon’s storage solution. That’s very low cost, and it’s very persistent. It also has some fantastic security features.

What we’re doing on top of that, because of the quantity of data, is rather than ingesting that into a database, we’re using Presto, which is an open-sourced query engine, to query that data in place. One of the software companies we use to provide Presto is the company called Starburst, and Starburst integrates with the Apache Ranger data security framework for Hadoop. Apache Ranger allows you to define fine-grain access to rows and columns on a per-user basis, and that’s really where we implement our security model.

So, you can secure those cells, if you will, at a very granular level, and that’s what gives you the privacy and security?

Jarvis: Exactly.

Describe how you see machine intelligence applied to data?

Jarvis: What we’re looking to do is really two types of artificial intelligence, machine learning, on top of our data.

The first is operational. Things like: Will your patient miss their appointment? That’s important, because if they’re very likely to miss their appointment, then you can remind them to come along or plan to use the time if they do not turn up for the appointment. That’s relatively easy with AI. You do have risks of false positives and false negatives with operational AI like this, but the consequences are not so severe as with the other type of AI, which is trying to predict people’s medical outcomes in some way, which drugs are best for them, what conditions might they be susceptible to.

There are still questions as to the best way to adopt this style of AI in healthcare. Do you constantly learn, which means the algorithm’s changing all the time, and therefore what you assured last week is different to what’s in production this week?

We’re working with some partners to test out what that would be like. It takes a long time to get the ethical approvals and get the data in place, but the consequences of being able to do it properly are immense. The amount of time that you can save on clinical staff, the effect you can have on patient outcomes, is just well worth the effort you have to put in.

The last 10 years have been about lowering the cost of accessing data, getting it to where now we can apply machine intelligence and artificial intelligence to really drive organizational value and patient outcomes. Can you share how EMIS has changed over this time? 

Storing hundreds and hundreds of terabytes worth of data is trivial now. We’re not focusing our effort on doing that. We’re focusing our effort on getting the data in a high-quality form, getting the machine learning algorithms ready to go rather than on a kind of low-value storage and access of the data. So, we use the Hudi Project, which is for Hadoop upserts. We do upserts on our data using that technology. And that’s worked very well, and we’re contributing back to that.

We have to act in a real-time way, because if your patient data is updated by one care professional, another one needs to be able to see it quickly in case there’s an action that’s required.

The cloud in general, but AWS specifically, has a large ecosystem that attracts start-ups, innovators and helps build out an ecosystem that can solve problems. How has that helped EMIS? 

Jarvis: There are two ways AWS have helped us. Clearly the technology is good, and a lot of it is easy to access, if you’re an engineer. But what I have found with the healthcare team in AWS is that they’re very happy to help with funding of things, supporting of things, understanding our architecture so they can improve upon it, or review it. Also introducing us to partners and helping our partners help us. That is important. It’s not just about a technology solution; it is about joining that all up together to provide us a service.

Watch the complete video interview below, and be sure to check out more of SiliconANGLE’s and theCUBE’s coverage of the AWS Public Sector event. (* Disclosure: TheCUBE is a paid media partner for the AWS Public Sector event. Neither Amazon Web Services Inc., the sponsor for theCUBE’s event coverage, nor other sponsors have editorial control over content on theCUBE or SiliconANGLE.)

Photo: SiliconANGLE