Data Literacy

Fancy getting a private blood test? I did . . .

This post talks about my health in a somewhat vague way because the specifics aren’t important here. I want to preface the whole thing by saying I’m absolutely fine and you don’t need to be worrying about me.

For the past couple of years I’ve been dealing with an underlying health concern. It might be nothing, but it doesn’t feel right to me, and after a particularly bad time in late 2018, I asked for a referral to a consultant. Fifty weeks later, the appointment came through. 

Once I got a date for the appointment, I started collecting data. I used a (research-led design) health app to track my mood, my pain, my triggers, and my symptoms. But I also sent away for private blood tests. 

There are lots of companies now which will run a range of screening tests on your blood, spit, or whatever else you’d like to send off in a biohazard bag through the post. There’s also a lot of debate about whether this is a Good Thing or not. 

To me this is a question of data literacy. 

With many long-running conditions, healthcare providers expect individuals to collect data. I have spoken about keeping a note of Athena’s allergies before:

I recorded this data in a little leather notebook I wrote in every evening. The app I use to record my current symptoms is just a more advanced version of the same thing. Both will help a physician to make decisions about treatment. The rub is that the vet and the consultant have been trained for years to make sense of that data, whereas I, as a non-specialist, might draw the wrong conclusion. What is meaningful information, and what’s just noise?

We all know the danger of googling symptoms, such as that time you had a bit of a sore neck and then you discover you have neck cancer and you should get your affairs in order. Extremes are noticeable and attractive. We propagate stories such as that time an infra-red camera scan detected breast cancer, because they’re interesting – and because they’re unusual. I’ve seen lots of hot patches on people using infra-red cameras, and its never once been a serious underlying condition. Because as a screening tool, its not that useful. 

Screening tests are problematic. We have two measures that are useful when we think about screening: sensitivity and specificity. Let’s imagine we’d come up with a blood test that aimed to diagnose whether you were in fact a bit of a dick at times. A highly sensitive test would read ‘positive’ for everyone who had ever been a bit of a dick. A highly specific test would read negative if you had never been a bit of a dick. The issue is that sensitivity and specificity are not always related. You can have a highly sensitive test that rightly tells all the dicks they’re dicks, but if it has low specificity it will also tell a lot of people who are not dicks that they are, in fact, dicks. And this will upset a lot of nice people. Conversely, a super specific test will never frighten the good people with a wrong dick diagnosis, but it will also tell a lot of dicks that they’re good people. We want both sensitivity and specificity in our tests. 

Alongside the issue of how accurate the test is, there’s the question of ‘do you need to know?’ Here’s a great thread on the risks of DNA testing for cancer here by Rachel Horton. One of the lines that stands out to me is: “The scarier a result, the more likely it is to be a false positive”

The human body is not something that can be easily categorised. Our biology is a spectrum. Not only may a test be wrong, but you might just be tootling along quite happily with any number of potentially scary numbers attached to you. You may have a perfectly benign cyst in your brain. You may just live with low iron levels. In fact, many GPs are incredibly concerned about how these tests can worry people for no reason. 

Just like you’ve got that weird fleck in your eye, or the way your finger bends at an odd angle, there are lots of little oddities in your body that are perfectly fine. The body has such a great ability to just manage with so many things that don’t quite fit the concept of the medically fit person. 

So with all these sensible caveats in mind: why have I paid for the tests? Like I say, this to me is a data literacy issue. There are rightful concerns to these tests, and healthy people should absolutely not spend money on this unless they really enjoy looking at meaningless numbers (hey – that’s why I have a FitBit, no judgement). But they did give me more information. Specifically that there was nothing obviously wrong other than being highly deficient in Vitamin D like all Scottish people. 

As a semi biology-literate person, I did deliberately select the tests that would be likely to be affected by the health concerns I have, and deselected a number of the tests that I did not think were necessary (but that the company were keen on pushing). I also have a reasonable guess at what’s wrong with me from my GP, and that was later agreed by the consultant.

Personalised healthcare is fast becoming a ‘selling point’ for a lot of services. I don’t think we can stem that tide, so we need to think about how we create more data literate people in all sectors, so others can think critically about what’s being offered them. But health is also scary. The health of our loved ones is particularly scary. Can we ever really be objective or analytical when it comes to health? Is that what these companies are preying on?

I don’t have the answers to these questions, but experiencing it myself has given me a new perspective. And I wonder if this is something we need to start teaching simply as a matter of course: how to deal with large amounts of your own personal data. Probably.

Oh – and after I went to the consultant? I was called to the GP for yet more blood tests.

If We Should Dress for Sun or Snow

Despite feeling pretty good about my work-life balance last year, I’ve been a little humbled by 2019 so far. My personal life has needed more attention than my work life, and I’ve been feeling guilty about shifting the focus.

Before Christmas I got very into the Groundhog Day musical soundtrack, particularly If I Had My Time Again, which is my new favourite shower sing-along. I was also thinking a lot about academic workload last year, and how the varying pressures of the academic role can be challenging.

Despite feeling pretty good about my work-life balance last year, I’ve been a little humbled by 2019 so far. My personal life has needed more attention than my work life, and I’ve been feeling guilty about shifting the focus. It’s been difficult to keep on top of things, and I hadn’t quite appreciated how much I’d let things creep into the evenings.

There were two articles recently that my mind kept returning to. One is Dr Anderson’s widow speaking out about academic workload, and this article about email’s influence on workload. Particularly on Monday when I was attending an Echo 360 community meet-up about learning analytics.

I had good reasons for wanting to go to this community meet-up. I’m interested in analytics, and I’m the PI on our university’s evaluation project so a little networking is always valuable. I’m also in the rare academic position of having some spare money floating around so it all seemed worthwhile. Except there was a very west-of-Scotland sounding voice in the back of my head wondering if I’m worth spending that money on. Who am I to go to That London to talk to people? Shouldn’t I be slaving over a hot laptop?

On the other side of this, I’ve also spent a little bit of my evenings this week working on a Shiny app. Now I want to emphasise that ‘a little bit’ in this context literally means five or ten minutes here and there when an idea comes to me, but it’s still very much useful time. And yet I’ve been frustrated that I haven’t been able to spend more time on it.

A couple of months ago I had a devil’s advocate style debate with my good colleague Ian about how much these kind of extracurricular activities should contribute to our CVs. We kept circling back to how much the open science and open data analysis movements favour those people with the spare time to dedicate to this kind of work. If all your work is on proprietary data, you maybe can only contribute to things like a github repository in your spare time. And if when you get home you start doing the childcare, or can’t get away with not cleaning the house because you prefer to spend that time tweaking a package. What if all your hours out of work are spent on other tasks, and when you have that lightning moment of “ah – I should use enquo()!” you can’t immediately go to your laptop to check it out?

There are many people much busier than me who manage to contribute way more than me. Those people should be applauded. And we should definitely still value the amazing resources people put online. I think it is our responsibility as academics to support ourselves (and our managers too).

All this is a round-about way of saying that having a little bit less time to make-up for my business has highlighted to me how very important it is to protect time for the things that are important in your work. During one of our protected analysis times today I started a new package which I hope will be able to be incorporated into a shiny app I’m planning for our students. Tomorrow’s my first Writing Friday since before Christmas.  This is the way to do it. And yes, my emails have been slipping in the mean-time. Let ’em.

We should believe we are worth the time.

(And also I managed to go to work today wearing two different earrings and no one pointed it out. That’s not relevant but it amused me greatly.)

Naughty and Nice

A small Christmas blog on the ethics of being overheard . . . he’s making a list, he’s checking it twice . . .

Amazon have put all five seasons of Person of Interest on Prime. Person of Interest is an amazing exploration of what it might cost humanity to create artificial intelligence, and its beautifully prescient given Amazon’s recent Alexa data breach where a user was able to access another user’s recordings.

In my book (which if you’re looking for a last minute Christmas gift, do check it out) I talk about how we might end up studying personality through artificial intelligence, and the ethics of how we might consider this data use.

I’m delighted with my Christmas present of Person of Interest. I cry the whole way through this show. It is amazing. But I also have an Alexa sitting in my house, and a Google phone. Occasionally my phone flashes its screen, saying “I didn’t recognise your voice”, much like Athena’s ears prick when she’s snoozing and hears me get to my feet. Do I need to listen for you right now?

On the other hand, in 2018 I’ve also had to balance the issue of not having ethics committee permission to share sensitive data and the challenges that has caused for making my research open and reproducible. I am proud particularly of this repository which will be elaborated on in a publication next year – how we can be reproducible when we’re dealing with data that should be confidential. But yes, privacy is a challenge.

And it’s a very strange conversation to have in December. He sees you when you’re sleeping, he knows when you’re awake . . .

Recently, I was asked how old I was when I understood about Santa. It actually ties in to my first experience with religion. I was raised without any religion whatsoever, and when I got to school I was introduced to this whole new concept. That my own mind and actions were not my own private space, that someone or something might be watching. I made a deal with this ‘God’ (who I pictured as Danny Devito in a toga, I do not know why). If I was very good, he would reward me with a hotdog on Friday at lunch time. As one of the Mac kids, I was always at the middle of the lunch queue and the hot dogs were always gone. So I was very, very good for a whole week. I did the praying. I was kind. And on Friday  . . . there was no hot dog.

The only other experience I had with religion was one of my grandfathers who had cryptically said “Any God who doesn’t want me isn’t a God I want to believe in”, and at the age of 5 I sanguinely accepted this logic, and decided the lack of hot dog meant God had no interest in my soul. That Christmas I tried this logic again, and created my perfect toy (a My Little Pony toy of my favourite character – except there would be movable bits). Santa did not come through.

We teach morality to children with the idea of oversight. Perhaps not entirely, but ‘being watched’ is a large component of how we learn our own moral frameworks. The Good Place has made an excellent TV show exploring the concept of being constantly observed (and measured). It’s probably not a coincidence that we’re interested in these stories right now. But it’s also not a coincidence that I got thinking about this after realising I knew more than I wanted to about my new neighbours.

Ultimately I think data collection and analysis is an organic process, and it’s very hard to draw a line over ‘good data collection and analysis’ and ‘bad data collection and analysis’. Amazon absolutely should not be sending clips of audio to a random stranger. But should I hear random snippets from my neighbours’ lives? How often should we accept being ‘overheard’ as a price of being digital neighbours?

I don’t have an answer for this – or even a reason to blog about it on Christmas eve. I just think it’s a very interesting question.