One of the big issues faced by all players in this sector is the lack of evidence for what we do. We are reliant upon overseas studies and hope that their recommendations work for our population. We know that this often falls short of the mark, but what choice do you have when there almost no local studies?

Its hard to get funding (read there effectively isnt any) in New Zealand for any veteran cohort other than the Vietnam veterans, whether you are talking population level care or research.

We all therefore argue from either anecdotal experience or opinion (expert is debatable). That needs to change. 

We are proposing to collect data - across a wide range of sources. That data will inform our decision making moving forwards. We also propose to make it available to researchers and the media (with appropriate controls of course).

In order to have veterans give us their information we need to do a few things;

1. We need to guarantee its security.

2. We need them to be able to see what its being used for.

3. We need them to be able to decide what will be used and for what. No big black box that you just dump data in and hope for the best.

What do people want?

Veterans want to be able to store their own data, share it how they choose to, and be able to access their data when they need to. Tell their story once and then point providers to that information.

Researchers want to have a reliable dataset that is not too difficult to work with. That is able to accommodate different data models and different study designs (prospective and retrospective).

No Duff wants business intelligence to guide the care decisions and also the advocacy  - evidence based care and change proposals based upon real evidence that has relevance locally.

We are not in Kansas anymore

We are proposing to gather information directly from the veterans themselves and from their medical record. Add to that published scientific / academic literature, Government records, and other information sources and you have this messy mixture of data which is hard to draw anything from. But there are two even bigger problems here.

The first is the technology problem. Traditional data storage is in SQL databases. They are great for storing huge amounts of data in tables when you already know how that data is connected. But they are really bad at storing data where you don't know the relationship between data points before hand and that is the really important part here. We actually don't know what we will be looking for - that's the point of doing the research, to look for what is happening. 

So we need an alternative that allows us to work with lots of data, to also to look at how that data is connected, but without knowing how the data is connected before we start building the data store.

Fortunately the solution already exists - its the same technology used by many social media providers and by many complex data systems. Its called a graph database for the geeks out there in case you want to look it up. It offers huge advantages over traditional datastore technologies.

The second problem is our patient population is somewhat paranoid (understandably so), therefore data protection and transparency of use is paramount. It is important that our solution then has encryption down to the individual data points to prevent unwanted access (hacking), and also control and trackability of data use down to the individual datapoints (transparency and use control).

We have looked at a number of possible solutions and are confident we have found something that will allow us to provide a safe and effective data store. So we are now into the design stage for the project.

Imagine what could be achieved if we knew what the veteran outcomes actually are - not just the mortality data (although even that doesn't exist at present), but down to the social, employment, financial and family outcomes. We know anecdotally that 1 in 5 have problems - but what happens to the rest?

Do those with post traumatic growth actually go on to do better than average? Do they live longer? Do their kids do better learning from their parents? This project isn't just about finding out the bad stuff to make more effective claims, its about what is actually happening to inform decision making for everyone. 

If we can build enough information we may be able to do predictive analysis. That means if we have enough information we can work out what factors predict a less favourable outcome or a better outcome. We could then look for people whom have those less favourable factors and suggest ways of reducing the probability of the bad outcomes. Imagine being able to intervene before the veteran develops a cancer - through better screening and surveillance.

We know these systems are capable of doing this - it is the same technology that ran the intelligence systems on operations so we have seen it in action. We just apply the same approach and bend it to something to benefit the veterans.

Data is essential for making informed decisions. More than this, in a situation where resouces are constrained, its unethical not to make informed choices and get the maximum benefit for the community as a whole.