A different year
This year’s intern scheme at Health Data Insight was intended to be different but we obviously didn’t intend to run it in a global pandemic. COVID-19 meant changes to the planning and running of the internship, and here we reflect on how it worked out.
Previously we have offered a range of intern projects, each undertaken by one student with support from their supervisor. Interns worked at the same time but individually.
For the 2020 internship we had already decided on a different approach; six students all working on one project. We felt that the range of skills required to pull a whole project together would be an interesting challenge for the group, and that interns would learn a wider range of skills from working as part of a team.
…as a mathematics student I usually don’t have a lot of chance to do a group project like that … we come from different backgrounds so, like, we can help each other specialise in different things so we work as a really good team. That is something that I didn’t expect but I really enjoyed about the internship. (Aisha, 2020 intern)
Face-to-face or virtual?
The Internship application and shortlisting process went smoothly; we used Mailchimp to communicate with candidates and Eventbrite to schedule interviews. This gave students the choice of selecting a convenient date/time and whether they wanted an in-person or videoconference interview (and also allowed them to reschedule, without having to contact us). This was the end of February, and we were able to interview all the students, either in person or online depending on their location, before the outside situation began to change.
It was during the process of making offers that COVID was spreading more widely and we began to discuss the options for the scheme and whether it was viable to run. The plan had been to have the interns working from the HDI Office space but it gradually became clear that this may not be possible. If we were going to run the internship at a distance we needed to be clear how we would communicate with the interns both before and during the internship. This created some tension in the planning meetings leading up to the interns joining us, as some staff felt we should provide the interns with much more detail than others.
We used Trello during these regular Zoom meetings as a way to collect and share the different opinions, to discuss them and to reach a resolution that everyone was happy with. This use of Trello to “display thinking” of participants was also key to developing teamworking in the virtual team that would be supporting the interns.
The Trello board used by staff to share ideas during and between internship planning meetings.
Setting the brief
The project outline was to ‘build a synthetic data service’. The ultimate aim was to for the interns to provide a system where data holders could request a synthetic version of their confidential data. In setting up the project, we had to balance giving enough information to interns so that they could be successful in the task, but also allow them to be involved in that specification.
I’ve really enjoyed thinking of the product as a whole from conception to completion. (Rebecca, 2020 intern)
Rebecca’s early visualisation of the project brief
It was challenging but because we had this aim and very clear goal set for the project it was so enjoyable and very rewarding. (Lulu, 2020 intern)
Virtual data access
Medical datasets are extremely valuable resources for research aiming to improve patient outcomes. However, because these datasets include real patient information their identity must be protected, resulting in a lengthy process to gain dataset access permission to be able to perform research. The reason we chose to focus on a synthetic data service was to improve access to these sensitive data sets. However, where we would normally have set interns up to work with the real data, virtual working meant this was not possible. Our team found a medical data set for malaria that had been made publicly available and that was of a level of complexity that would match the sensitive data that we planned to work with.
The work was really interesting … I got the chance of implementing AI algorithms based on research papers that were published not long ago. (Evonne, 2020 intern)
Teamwork and tech
Time was devoted not only to ensuring that the students had access to the technology, software and data systems that they needed to do the project, but also to communicate effectively with each other. Key in this, was our intern team lead, who had worked with the previous year’s interns and provided a first port of call for any queries.
We felt it was important that the communication channels were clear and encouraged team interaction. Setting this up started with an introductory morning with the whole HDI team to get to know each other. We used zoom break-out rooms to form smaller pairings so interns could talk more informally to each other and us without the pressure of speaking in front of the whole group. Following on from this there was a daily ‘stand-up’ Zoom meeting with the intern team lead each day to discuss progress and next steps.
Interns in a daily ‘stand-up’ meeting with their intern team lead (yellow surround)
Aside from the regular meetings, we wanted to ensure that the interns felt they had enough freedom to explore the project and really make it their own while feeling supported.
I got to try out so many new and different things… in a professional setting (Faiz, 2020 intern)
We suggested that the students organised themselves into groups to work on different parts of the project depending on their skill set and interests. They were also encouraged to move between the groups and take on different roles:
I worked across the AI algorithm team and the evaluation team… analysing, cleaning and formatting the original data… in the middle of the project I was trying to understand the federal privacy and how we could apply that to our algorithms… towards the end I was helping with producing the evaluations for the synthetic data and I also got to design the logo for the project. (Lulu, 2020 intern)
Lulu’s designs for Syndasera and the results from the preference vote
We used Trello boards for sharing information about the project, Slack channels for queries, Google drives for outputs and towards the end of the project the interns setup an AWS server to trial secure uploading and release of data.
The interns project board
At the end of the internship we scheduled a ‘presentation day’ where the interns presented their progress to HDI staff and Board members.
…this internship has made it very clear to me that I am interested in statistics and health data (Lulu, 2020 intern)
Whilst we all agreed that we would have preferred to run internship in the office, this distance internship worked out very well in the end. And due to the pandemic, our interns ended up spread around the globe, from China and Malasyia, to Portugal, London and Durham, although to be honest, we wouldn’t have known if they hadn’t told us. We are delighted that the students got so much out of their time with us and we look forward to another international internship next time.
You can read all about their Synsadera project here.