Here at Seldon, we are deeply concerned about the impact of COVID-19 on people’s lives. The people on the front line of healthcare and research are the true heroes of this time – and everyone has a part to play to support them and flatten the curve. We believe AI has the power to offer transformative solutions to the immediate crisis and build the post-pandemic world.
We’ve seen an increase in initiatives using open datasets and open-source software from organisations across the world. Some high potential initiatives on COVID-19 include open research paper dataset made available by the Allen Institute for AI as well as the data repository by Johns Hopkins CSSE. Both are examples of how open-source data science projects can play important roles in keeping the public informed, influencing global policymaking and helping to combat the spread of the disease. Other interesting projects include an initiative between NHSX and Public, who have announced £500k of funding (on top of a £25k TechForce grant challenge) for innovators who can offer digital solutions to support those self-isolating because of coronavirus.
We want to support the machine learning community as much as possible. In just the past week, we’ve seen an 11% rise in usage of Seldon Core, a signal of a continued appetite for ML technology and the part it can play in creating solutions at this time. We believe the most powerful solutions come through cross-functional collaboration across domain experts from various backgrounds. We want to encourage this type of collaboration and empower the great initiative that data scientists have taken to dive into some of the open datasets.
There are over 1500 projects on Github tagged with COVID-19 (update 2020-03-30). Open collaboration, combined with the power of epidemiologist domain expertise and MLOps infrastructure tools, can have great contributions to ongoing discourse and research. In this spirit of collaboration, we’re opening up the hive of expertise in our open source community, made up of thousands of members across the world. If you, or someone you know, needs assistance at this time, let us know by emailing [email protected] or join us on Slack.
As a starting point, our ML Engineering Director, Alejandro Saucedo, has put together a practical example showing how data scientists would be able to deploy their model at scale – in this example he uses Seldon Core and our powerful Alibi functionality in order to explain the text predictions from a COVID-19 research paper classification model trained on the Allen Institute for AI dataset. You can read the full tutorial on Github.