After six weeks working within the PPSI team at Swansea University my summer internship is coming to an end. While certainly hard at times, I still find myself wishing it were longer. Throughout my undergraduate degree in Physics and Philosophy, I found myself more and more intrigued by data science and its real world applications to make people’s lives better and easier. I taught myself enough Python to ‘get by’ but before this had no real experience of writing code for applications to be used by others. With my sights set on post graduate research in computationally heavy healthcare related fields this internship seemed the perfect opportunity to develop my skills and understand how research teams work in the real world.

The main project I worked on was a Python library called Cognitio. The goal of Cognitio is to create an environment in which to run entire machine learning model pipelines in interesting and repeatable ways. This was quite a daunting task at first, I had previously never had to worry about the structure of an entire library or how to design code in a way that allows updates like parallel processing in the future. It was certainly a struggle at times, but Marcos was extremely helpful and working in a very skilled team like this was the perfect place for me to expand my knowledge. I can now say I am much surer of designing programs well and I’m much more confident in my ability to understand and debug larger code bases.

Outside of the technical aspects, I found working on Cognitio quite exciting due to the potential it has for making much needed research a lot more accessible. Bias in artificial intelligence is a hot topic at the moment, and something I am particularly intrigued by. Especially in the context of healthcare, it seems of the upmost importance to ensure new technologies can provide fairer outcomes for everyone. It is here that my work on Cognitio has the biggest potential in my eyes. The performance of models can be tested against different categories or demographics to easily show where models, or the underlying data, may be showing biases. What I really took away was an insight about how you can start to tackle large, industry-wide problems (like AI bias) by designing your models and experiments with them in mind from the start.

I also contributed to a research report conducted for the Samaritans charity on self-harm and suicide imagery on social media and the issues surrounding that. I am very aware of the power social media has over the lives of young people in our society, and find the implications and impacts for mental health very interesting to think about. There is also lots of intersection with topics like online censorship and online safety legislation which I found very interesting to think about through a different lens. This was a great opportunity to combine my pre-existing interest with new information, data and ideas.

The best part about this report for me was the chance to contribute ideas, based upon real data, about how social media companies could limit harm to their users in new and better ways. Even if many links down the chain, being able to contribute ideas that may have some influence is something I really enjoyed. I learnt a lot about mental health in young people and was sometimes shocked at the data - especially when looking at the youngest members of society. More than anything this has made me more motivated to carry on doing research within this area. I can now really see how data science can improve the interventions and treatments healthcare provides, and how this can lead to measurable improvements in people’s lives.

I have learnt a lot from this internship, not only about mental health data and machine learning but also about how research teams’ work. I wish I had more time to continue working on my projects here but I’m aiming to continue contributing after I leave. It has also given me a lot to think about in terms of my plans for the future. Going into this, my plan was to study a computational neuroscience PhD after I finish my masters next year. I had never previously considered epidemiology as a career path but it’s something that feels very appealing now, I think I’ve got a lot of researching to do to see exactly how that will look for me but I’m excited at the prospect.

I’ll miss working with the team here (and the baby seagulls outside the meeting room) and extend a massive thanks to everyone that’s made this experience so positive. I hope I’ll have the chance to work with some of you again in the future!