The rapid acceleration of Coronavirus is not lost on any of us. It continues to affect how we work, socialise and go about our lives, and will do for some time to come.
KJR is a proud member of the newly established AI Consortium Pty Ltd. Along with intelliHQ, AIkademi, Max Kelsen and 9Points, we are working together to contribute positively to the innovation and integration of artificial intelligence across the globe. The first stop – tackling the question “Can AI help in the fight against COVID-19?”
Maybe not the first thing typically associated with a pandemic, but the reality is that many clinicians and healthcare professionals work in conjunction with data science and AI modelling to help inform medical decisions. Through the collation, testing and modelling of data, predictions can be made to guide decisions and help reduce the stress of COVID-19 globally.
On the 9th of April, as is the current norm, we gathered over a video conferencing platform with an exceptional panel of experts to discuss the very real implications of AI in this virus-led world.
This panel featured experts in their field, who are all active in the fight against COVID-19:
- Jeremy Howard, founding researcher at fast.ai
- Dr Brent Richards, Medical Director of Innovation and Director of Critical Care Research at Gold Coast Hospital and Health Service
- Dr Sally Shrapnel, an AI lead on a current COVID-19 international data collation project
- Professor Kerrie Mengersen, Director of the QUT for Data Science
- Lucy Shinners, PhD candidate focused on exploring healthcare professionals’ perceptions of AI in care delivery
You can watch the full webinar right here. Or, if you’re suffering video fatigue, get a recount of key themes in the written word below.
Collaboration in times of need
Brent: “We’ve got so much going on that frankly we need as many hands-on-deck as we can get. The more we can get people who understand how to work with data, how to compute it and then turn it into something that is understandable by a clinician is at a great shortage… we’ve needed it more than we’ve ever needed before.”
Lucy: “Down at a clinical level, we’re always playing a game of catch up, for clinicians their focus is the patient in the bed beside them … Clinicians know they’re creating data, but they don’t understand just how much value we can get from that data science… So there’s a lot of work that needs to happen.”
Sally: “There’s a bit of scepticism about whether data explainability techniques are really explaining what we want them to… In terms of bringing it into the COVID discussion, I think it’s important to have domain experts involved all the way along. At that point you can determine how to apply the data to the question they’re asking.”
Jeremy: “Who’s an expert on the use of cloth masks for COVID-19? We become the experts and studying data is a great way to that, but we also need to apply it to the situation when we make decisions.”
“The American Statistical Association guidelines state that we should not be using P-values as the key basis for health policy decisions. What we should instead have done is use the real word ‘evidence’ to make decisions, like asking if there are any signs of human to human transmission, before relying on studies.”
Brent: “It’s been good to look back and use old data models to understand what we normally do… If we are running into a data shortage, we can say ‘well, there are other ways of doing this which seem to be working well in other situations’ … There’s some very big ongoing uses of data which are part of the preparation as much as the understanding.”
Data for the public’s needs
Sally: “One of the things we’re working towards is putting together a data dictionary for these kinds of problems so we can release it publicly. As more people come onto the project, they need that quick start in knowledge to understand the domain and relevant data.”
Kerrie: “There are the scientific papers and theoretical work, and there’s also the interpretation of that by data scientists as well as clinicians. More than having a register of data scientists, it would be really good to have intermediaries who aren’t on the front line, who have time to talk to the public and answer questions.”
So, is AI helping or hindering the fight against COVID-19?
As the world responds to the virus, AI and data modelling have already begun playing a major role in informing decisions. That said, the importance of being open-minded in professional settings to interpret this data is more relevant than ever. Working collaboratively with professionals from different industries and with different perspectives will ensure crucial outcomes in preventing the outbreak from further escalating, particularly in the healthcare and data science industries.
Utilising AI and data modelling to project potential COVID-19 pathways has been limited with the topic’s growing database. Overstepping this has involved a mixture of sorting through the noise and looking back at previous models and databases such as the MIMIC-III Clinical Database Demo to see what worked in the past and how this can be applied to make rational decisions. Being able to provide the public with objective, unbiased information is essential, which is why initiatives such as ECMOCard have released data publicly for those interested.
Our panelists certainly confirmed that AI and data science are playing an important role in the COVID-19 response. However, we must remember that AI experts alone won’t end this pandemic – we’ll need to collaborate and utilise our collective brainpower to establish the right steps to take.