Analytics and AI in Healthcare

By Steven Parrish, CIO, Taranaki District Health Board

Steven Parrish, CIO, Taranaki District Health Board

Healthcare is a data rich sector and there is immense value that can be derived from it, by turning data into information and insights to drive an understanding of the sector and the immense changes that are needed to ensure efficiency and effectiveness for the ultimate benefit of the patient.

One would think that this is obvious and when you look at other sectors, the benefits realised through the use of data analytics and now artificial intelligence (AI) is prevalent. Even overseas in the health sector the use of data through analytics and AI driving insights is seen as the norm and just part of what they do. So what about New Zealand?

While I have only been in the NZ health sector for just over 18 months, what I have seen shows that we have a good foundation, previously holding a position of global technology leadership throughout the 90’s and early 2000’s. However, with a lack of significant targeted investment our foundation has not been expanded on. This does not mean that nothing has happened but it has relied on the ‘kiwi’ can do attitude and pockets of innovation.

The Health sector in New Zealand has a number of things in its favour when it comes to the foundational components. This includes the unique identifier that is the National Health Identifier (NHI), significant uptake of electronic systems within Primary Care, home grown Health IT companies that compete on the international stage and a relatively small population base.

Some of the limitations to advancement in Analytics and AI in healthcare and particularly the public sector include the proliferation of paper based processes and therefore data, lack of agreed standards, increased reporting requirements from central government, funding models that focus on activity and not outcomes or innovation and the challenge of getting skilled and experienced staff.

With the proliferation of Analytics and AI based products and services hitting the market we need to provide an ecosystem that allows Health to take advantage of these products and services to realise the potential benefits that are proposed. The benefits especially around Analytics are well documented and researched especially when looking at the international experience. AI benefits in health are often theoretical and while have a sound base need to be proven within a research construct to add a level of validity. I say this, as within the discipline of medicine and health, research is the foundation of best practice and without this, there is scepticism of value.

To start the focus on Analytics and AI I believe there are some key learnings from my experience that that can be applied.

1. Minimise manual data collection

In order to get the most out of both analytics and AI first you need data and more specifically electronic data. Healthcare is traditionally very paper based and this can limit the ability to analyse the data to obtain insights or run the AI algorithms. In the collection of data it should be an outcome of the workflow for the user and not a specific focus so that they see value.

2. Static reports vs. data views

We need to move the conversation from provision of static printouts and reports to one of dashboards with drillable data views. This is a conceptual conversation that in my experience key users do not understand as they fall back into what they know and what they are used to. The best way to move the conversation is to do a proof of concept with a key content type and user group to show what is possible. This then provides a demonstrable example of what the difference is between reports and data views.

3. Data extracts vs. Virtual integration

Historically we extract everything from a source system, put it into a large data warehouse and then look at running the analytics across it. There is still a place for that, however, in order to minimise storage and costs there is the ability to integrate virtually with source systems and access the data as needed when a data view is being accessed. One needs to be cautious about performance of the source systems with this so a hybrid approach often is required.

4. Strategy

Organisational buy into any project is key to its success and analytics and AI is no exception. Development of an Analytics and AI strategy needs to align tightly to any organisational strategy to ensure visibility of value to the Executive and Board. This allows for conversations around value and justifies the projects existence. Where organisations see the value you may find that one of the pillars of the organisational strategy is specifically about data and insights, which ties in Analytics and AI.

5. Governance

Once you have a strategy and agreed way forward there is a need for efficient governance. Governance gives the oversight and direction for the organisation and helps you navigate through each challenge as they come up. Membership is key and having senior members of the organisation on the governance board will assist in moving things forward. If membership is delegated you should ensure that the delegate is communicating up and representing broadly the area they represent and not using it as an opportunity to benefit their own agenda.

6. Data and Information Management

The discipline and principles of Information Management are key to getting the best out of the data. Data quality is important, as this is where data confidence is founded. The data does not need to be perfect however, one should understand the level of confidence you have in the data so that this can be considered as the data is used to drive insights and decision-making.

7. Agility

Key to everything that we do should be the ability to be agile. This could be founded in Agile Frameworks and DevOps however in broad terms one needs to be able to change direction, deliver regularly and often and fail quickly while learning as you go. Keep the user involved and adjust as they increase their knowledge of what is possible and therefore change the expectations of what they see as valuable.

With the continued reporting of Healthcare issues, cost overruns and the Minister of Health’s recent announcement of a Health and Disability sector review, healthcare in New Zealand should be looking at the use of Analytics and AI as one of the toolsets to drive efficiencies and cost savings while increasing the focus on patient outcomes and experience.