CEO of Metropolitan Health, Dylan Garnett, talks about the impact that Natural Language Processing (NLP) could have on preventative care and the healthcare industry as a whole and his insights on realising NHI.
What is your vision for Metropolitan Health?
We want more care for more people and to unlock financial wellness, and that means bringing equality back to healthcare for all South Africans. One way we believe we can achieve that, is by focusing on the needs of the individual and putting responsibility for their health and well-being back into their hands.
We have developed a three prong approach to preventative care that starts with knowing your status, be it healthy or unhealthy. Once you know your status, own it and becoming responsible and accountable for your health. And lastly, improve it. I think it’s important to bear in mind that income and educational status don’t always determine an individual’s accountability for their health and so we are looking at tools and technologies that will help us achieve better health individually and collectively.
Metropolitan Health recently announced the first commercial application of IBM Watson technology in Africa. What was the driver for bringing NLP technology to the local healthcare environment?
We are excited about the role technology will play in the future of healthcare. If we consider the kind of data we need in order to make real progress towards preventative care then we can’t just look at clinical data. We have to start looking at genetic, lifestyle and behavioural data too – and one of the challenges we’ve had is combining these data sets. The difficulty is that lifestyle data is in an unstructured format and linking that with structured genetic data and semi-structured clinical data requires NLP technology.
We saw a huge opportunity to utilise Watson’s capability to analyse unstructured data and use sophisticated algorithms to extract and link those datasets across complex patterns.
Watson will be introduced into your customer services environment to act as a ‘virtual coach’ to help Customer Service Agents in their interaction with clients. Is that correct?
We took the decision to start our journey with Watson in our customer services department because the first thing Watson has to do is learn the local healthcare language set which comprises two different sub-sets.
The first is the national language set so that it understands the South African use of English as opposed to US, UK or Australian uses. It needs to understand how we construct sentences and nuances such as the use of certain words in a particular order and the varying meanings. The second is the domain language set, which is the terminology used in healthcare that is specific to SA, such as ICD-10 codes which in some instances are unique to us.
In order to derive more value from the technology down the line it needs to start building up a body of language from both the domains’ perspective and a natural language perspective. Applying it in the call centre means it gets substantial exposure to the different language sets. We’re confident we will realise some immediate customer service benefits for our clients, and for ourselves, by applying it in the call centre. Our goal at this point is to use data to transform every customer interaction and deliver personalised care in line with our vision, shared with many other healthcare entities, of moving towards preventative care.
Do you have a vision for the use of NLP in a clinical setting?
Absolutely, once the language set becomes more mature, we will work with GPs and specialists to harness Watson’s unstructured language capability and complex processing to identify patterns in disease types like Diabetes. The technology will enable and facilitate better diagnostics and treatment paths.
Often in a clinical setting, a complete picture of an individual’s health can only be seen when we also consider the clinical narrative which is created during consultation with a patient. But patient records captured in an EMR/EHR system are typically template-based forms and the clinical narrative can’t be accessed meaningfully because it is unstructured. But if we could structure that data, we could gain many more insights and derive value from those insights.
We’re committed to designing a solution together with doctors to understand how it fits into their workflow and how to derive maximum benefit from the technology within a clinical setting. Our goal is not to replace but complement existing processes.
We talked briefly about the fact that a lot of healthcare organisations don’t often know what questions they want answered from their data. Should organisations have a plan for their data before implementing NLP technology, or will they be able to better understand their business and clinical problems by using an NLP solution first?
I see it happening in parallel. When you start looking at how to derive value from your data it forces you to rethink how your current data is captured, structured, stored and what exactly it contains. One of the benefits of implementing sooner rather than later is to not spend a huge amount of money to structure volumes of unstructured data that have been stored over time, which is often the case.
At the same time, you won’t start off with a complete dataset either and you have to take it step by step. As the NLP tool matures, so will your data, and the payoff will be the insights you gain – particularly from the information stored as unstructured data – that helps you make improvement to business processes, quality of care and patient outcomes. You have to see it as a dual process.
Let’s talk about NHI. Metropolitan Health has released a green paper outlining the organisations’ position on the programme. One of the questions that remains unanswered, and for many is unanswerable, is how do we calculate the affordability of NHI? Does it need to be rolled out before we know if it is a sustainable solution?
Let me start by saying that as a country, we definitely need NHI and we need it to be successful. I keep reminding people that by and large we already have NHI – it’s called the public health system. It comes down to the difference between what is outlined in the government’s white paper on NHI and what we currently have which is a huge funding deficit and limited understanding of how we will redistribute funds and resources across the public and private systems.
Saying that, and the fact that we have to make it work, it’s not going to be an overnight process. To answer your question regarding affordability, I think we will have to start incurring additional taxes upfront purely because the investment deficit is massive.
I believe the key to NHI success hinges on a strong, stable partnership between the private and public sectors and that includes people like us who are sitting strong health capabilities, as well as hospital groups and right down to individual physicians. The bottom line is that if we want to make healthcare affordable and accessible to all South Africans it’s going to mean working together.
One of the priorities for the Minister leading up to NHI is improving the quality of care in the public system. What technology do you think they need to implement to do this?
If we could use another success story in SA, I think SARS is a good example to learn from. They understood that an effective electronic system was fundamental to improving their service but they took the “people, process, technology” approach, as we all should. It started with effective leadership, and then they upgraded the skill set of their employees, and then implemented smarter processes enabled by workflow technology. I think that the public health system should take a similar combined approach and I believe the Minister has started the processes by relooking at the management in hospitals. It perhaps hasn’t been a flawless overhaul so far but it’s a difficult process that will need refinement over time.
From a service point of view, it is a good opportunity to empower nurses in primary and tertiary care settings by upgrading their skills, both technical and occupational. In my mind, nurses will be the backbone of NHI and I think the Minister’s investment in nursing colleges was the right thing to do. It will take time to fill the backlog of skills and, even once qualified, it will take even more time for those individuals to gain the appropriate level of experience.
At Metropolitan Health, we’ve identified effective medicines dispensing as paramount to improving the quality of care. We’ve started working with clinics in a few regions to achieve better utilisation of the medicines spend and service in general. In total, SA spends R8bn a year on ARV drugs alone. I don’t think we’re getting value for money so the logical place to start for us is at clinics that dispenses medicine to build up a comprehensive dataset, improve patient adherence to their drug regime and implement the tools that enable a 360° view of the overall medicines dispensed, their monetary value and how to improve that and patient outcomes.
What can the private sector do to prepare for NHI?
We are starting to engage with policy makers and looking at the macro-policies and structures that will allow productive working relationships to form, without compromising the businesses we’ve built. Unless that happens I think we’ll end up with a lot of conflict in the system. An example is when the DoH tried to contract GPs and only 7% of the GP population showed any level of interest. We need to get some fundamentals right to enable NHI’s success and building trust between the government and the private sector, is a good place to start.
There are pockets within the private sector that can continue to say they aren’t interested, and won’t be supporting those policies. But what will happen will follow the simple premise that “he who makes the rules always wins the game,” so if they don’t come to the table or show willingness to participate, the rules may be made for them and that would be counterproductive for all of us which is why engagement from the private sector is important.
The first step is participation and showing the right intent. How do we bridge the gap between a public health system driven by service and a private health system driven by profit to create a homogenous healthcare system that provides universal healthcare access? That’s where we need to start the conversation.