Prevention is better than cure, so the saying goes, and I’ve been thinking about what kind of tools could be built to help individuals and organisations know when they themselves or somebody they’d like to help is at risk of harm. This really thoughtful post by Rev Dan Catt got me particularly thinking about depression. He writes about how he really understood that something wasn’t right because he looked at the data he regularly collects about his Twitter, Flickr, email and IM use:
What I was seeing was a change in my behaviour, a measurable mental state. And once Iâ€™d seen the numbers it made it easier to figure out what was going on.
Behaviour based apps that alert you (or somebody who can help) when you’re at risk might be a really interesting avenue. As Dan writes in the post, there are a few signifiers that you could use to spot a change in behaviour and it would be an interesting machine learning problem to solve them for different people. If you’re thinking about the NHS, the idea would be that promoting particular apps would be a way of avoiding people being admitted into the system in the first place. It might not just apply to depression — are there things that happen before older people have falls for example?
Another approach would be to look at bigger data. Most public sector organisations I’ve spoken to had no idea how much data they really had and didn’t know that even unstructured data can be analysed using tools like those Mastodon C have at their disposal. Once a pattern has been recognised (often using historical data from when things have gone wrong before), it’s not hard to set up a system that spots the early stages of that pattern and tells the agency it should intervene. This could be just as applicable to social services as the NHS.
And then there’s some really interesting stuff going on around new forms of diagnosis. This software that can detect Parkinson’s disease just from a phone call might just be one of a whole raft of new inventions re-imagining how we go about diagnosing conditions. And often, if you catch something early you can do a lot more to treat it.
Most of the time when we think about how to solve problems we have a cognitive bias towards solutions to the symptoms rather than the cause. But I think this kind of prevention engine has a lot of potential. It might sound a bit Minority Report but by being able to spot things before they happen we could save an awful lot of pain and money if we get it right.