Akshay Swaminathan (top) and Lathan Liou (bottom) are coordinating the writing of this article over Zoom, Google Docs and Slack.Credit: Lathan Liou
Although we are both currently medical students, we have spent a combined ten years doing data science in academia and industry, in roles such as head of data science at Cerebral, a tele-mental-health company in Claymont, Delaware, and a researcher at the Icahn School of Medicine at Mount Sinai in New York City. Our medical-school research involves working with clinicians and data scientists to deploy machine-learning models in health systems (A.S.) and developing epidemiological models to predict the genomic subtypes of coronary artery disease (L.L.).
As long-time friends and collaborators, we’ve often had déjà vu moments while talking about our research, finding parallels with our careers as data scientists. In particular, we both regularly find ourselves feeling frustrated by the limited use of standardized software tools in academia, which makes collaboration more difficult than it should be. These tools are commonly used in industry, so why not in academia? In this article, we describe a suite of tools that has helped us to make research projects more efficient.
Manual scheduling is tedious
Most of us have probably seen an e-mail that looks like this: “Availability: M: 1–4 p.m., T: 10–11 a.m., W: 9–10 a.m., 2:30–3 p.m., 4–5 p.m., Th: 2–3 p.m. and F: 10–10:30 a.m., 11–11:15 a.m.” and had to work out where our own availability overlaps.
The process becomes ever-more complicated as the number of meeting participants rises, especially if some of them are in a different time zone. The use of team-wide calendars with an open meeting-scheduling policy — such as Google Calendar or Outlook Calendar — enables anyone to schedule a meeting wherever a shared open timeslot is available.
If you’re trying to schedule a meeting with someone external to your institution, Calendly is a great free software tool that enables you to easily share your Google or Outlook calendar and allows other meeting participants to select a meeting time when everyone is available.
Get started by setting up a Calendly account for yourself and sharing your Calendly link when scheduling meetings.
The problem of multiple versions
Searching through our e-mail inboxes for files can be difficult. It’s particularly challenging during the preparation of manuscripts, when multiple offline versions of a Microsoft Word document are edited by different people and sent in an e-mail chain, making it hard to accurately collate everyone’s edits and comments.
Using collaborative real-time editing programs such as Google Docs greatly alleviates this problem. It ensures that everyone’s suggestions are incorporated into the same document and provides an easily accessible historical record of all the previous versions, in case there is a need to refer to deleted material — or even to revert to an older version. It also saves co-authors having to stagger the timing of their work, because Google Docs allows multiple people to work on a document simultaneously.
Similarly, for data, there are tools such as Airtable. Airtable is a good stepping stone between working with individual CSV files — spreadsheet-like files that can be used to tabulate data — and setting up a serious database. Meanwhile, GitHub is an industry-standard tool for code management. It includes functionality for code review and version control, so if conflicting changes are made to a piece of code, users are forced to address them.
Get started by drafting manuscripts in Google Docs or in Word in OneDrive, Microsoft’s cloud storage and file-sharing app, instead of using Word or a similar word-processing program locally. Share these collaborative documents with your teammates and encourage them to suggest edits and make comments.
If your research project involves code, create a GitHub account, make a private GitHub repository, upload your code and encourage your teammates to refer to the private repository for the most up-to-date version.
The difficulty of tracking progress
The process by which a research project moves from ideation and hypothesis generation to analysis and writing involves many moving parts. As research teams get bigger, so does the complexity of managing this process. We have found that software such as Notion (Trello, Asana and Jira are alternatives), a knowledge-management system organized in a series of relational databases, can help to break down a large research undertaking into more manageable chunks. For instance, we have used visual task-management tools called kanban boards for each manuscript with the following stages: concept, research, writing, revision and publication. Each task in a kanban board has notes, an assigned owner, a task timeline and, if relevant, links to other projects.
The benefit of using these tools rather than coordinating task allocation over e-mail or text message is that everyone is kept aware of team-wide progress on tasks, and information inequality is minimized.
Get started by creating a Notion account and using one of its templates for a kanban board to begin tracking your research project.
The difficulty of effective digital communication
Two other collaborative tools we’d like to mention briefly are Miro and Loom. Miro makes it easy to create a virtual ‘whiteboard’ that teams can use to, say, brainstorm projects by creating virtual sticky notes or sketching diagrams on a blank canvas. Loom allows researchers to easily create screen recordings showing step-by-step guides to navigating a task. It is the software equivalent of having a collaborator on hand to guide you through a workflow.
Get started by downloading Loom and recording a brief practice tutorial that walks you through computer-based research protocols.
Encourage the adoption of technology
Incorporating these tools into your research team’s workflow could lead to smoother communication and more efficient project execution. Just as industry-inspired retrospective or ‘retro’ meetings — structured discussions that allow teams to reflect on finished projects — improved collaboration in academia, these tools can revolutionize how research teams operate. By embracing technology and adopting best practices from industry, research endeavours become not only about searching for scientific truth, but also about ensuring that the route taken is efficient and reproducible.
Our final word of advice is that software adoption won’t happen unless there is top-down institutional agreement or bottom-up team-driven agreement — or, ideally, both. The whole team should be on board with migrating to a new project-management workflow. If there is already a system in place that works well for everyone, there might not be any need to adopt the tools that we have described. However, if a change is on the horizon, but some team members are hesitant to commit to it, securing the support of the principal investigator or another person in authority is often the most pragmatic approach to catalysing team-wide adoption.
Source link
