Students take on Wellcome Ideathon

The Wellcome Ideathon was an opportunity for small teams of students or researchers to propose data science solutions to tackle three urgent health challenges: mental health, infectious diseases, and climate and health.

Image

The four person team, consisting of second year PhD students Cameron Wilson, Emily Adams, Jonny Flint and Lara Johnson, made it to the semi-final stage, where they presented to Wellcome staff and guest judges.

The competition gave the Academy students an opportunity for interdisciplinary group work after their training year. They were tasked with developing a platform for long-term participant retention in mental health. They developed the front-end of a mobile application (called Discover Me) for adolescents and young adults participating in clinical research studies on mental health.

We are incredibly proud of the team, for coming up with the proposal, taking it on and making it as far as they did. Here, Lara Johnson reflects on what she took from the experience from start to finish:

------------------------------

We had a very intense three days at the Wellcome Data Science Ideathon semi-finals in London. We were pleased with how our presentation went on the final day but were sadly not selected as one of the three student team prize-winners. There were a total of 24 teams (nine student and fifteen researcher teams) at the event out of 100 initial applicants.

• The idea for the Ideathon originated 18 months ago, as an experimental idea for how to make applying for funding more fun than the typical process of writing grant applications. 
• I really enjoyed working on such a tangible project as a team. We had to write a proposal together, pick a team name (we chose “Iterative Insights”), articulate our values as a team (a requirement of the proposal), define our respective roles, brainstorm ideas for a solution and then implement those ideas (which involved coding). As the time window for working on the challenge was just two weeks, we also had to think about what was feasible to put in the prototype / demonstration to show our ideas.
• I learned a lot from the specific challenge we worked on. I already knew that attrition (and attrition bias – where the participants who drop out are systematically different from those who remain) is a problem in longitudinal research, and I was familiar with some of the statistical ways for trying to compensate for this. Over the course of working on our challenge, I learned much more about how much of a problem participant drop-out is, especially in mental health research, which typically has a drop-out rate of over 30%.  This got me thinking about what could be done during the research design – including through patient and public involvement as well as making use of new technology - to reduce participant drop out.  How can we as researchers make the process of participating in research easier?  
• In our solution, we distinguished between continual retention (keeping users engaged), threshold retention (identifying participants at risk of dropping out) and recovery (bringing participants who have dropped out back into the study). Our approach to continuous retention focused on developing a personalised research platform for participants through an interactive mascot and gamification (for younger adolescents) and using generative AI to generate personalised insights (for young adults).  Our approach to threshold retention involved developing metrics for predicting drop-out (which took into consideration individual characteristics, app usage patterns and satisfaction of participating in the research) to calculate individualised attrition risk scores, which could be used to assign a number of intervention strategies such as absence messages and embedding peer support workers into the research.
• I also learned a significant amount from the other members of the team. I learned a lot from Cameron on app development, Emily on user personas and people experiencing mental health issues and Jonny on longitudinal studies and from reviewing the data science literature on retention. Participant retention is similar to customer retention, a vital metric for businesses. There are established approaches and methodologies for measuring and predicting customer retention and loyalty. We used these as a starting point but sought to broaden this approach. We not only wanted to minimise drop-out but also embed an inclusivity aspect to minimise drop-out in certain groups (important for  avoiding attrition bias).  Given that this was in the context of health research, we wanted to ensure that all data collection (even implicit data such as app usage data) was transparent and explainable to the app users. Finally, we sought to make our proposed solution appropriate and ethical in a mental health context (for example, we wanted to ensure equitable treatment across groups and make sure there was no penalty or incentive for not engaging with the app).
• It was a good learning experience to hear the pitches of the prize-winners, both from an ideas as well as presentation point of view. It was interesting to compare the different approaches taken by different teams working on the same challenge. Hearing the questions that the judging panel asked gave a bit of an insight into the considerations reviewers have when reviewing grant applications.
• The in-person event included sessions on lived experience (which we felt very well prepared for thanks to our very own Jenny Sharma), inclusivity and presentation skills. The presentation skills session was delivered by the head of the Life Sciences arm of a venture capital fund that specialises in academic start-ups. She has heard about 2000 pitches for funding. She focused mainly on presentation delivery, and had us film each other and watch our body language.  This made me think about how we best pitch ideas for funding, which is different than delivering academic presentations.

A big thanks to all the team for being such troopers, and to our supervisors and Academy leaders for their comments and support throughout.