Computer Vision News - April 2021

accuracy: what are the chances for them to go on and develop some problem later in life? Is there anything specific that you would like to achieve, even if it would be difficult? Can you really reliably detect at birth which babies will go on to be at significant risk of long- term neurological impairment? That ’s really difficult. There are environmental factors. There are compound behavioral measures. You’ve got to marry many high dimensional datasets. You’ve got real problems trying to robust and generalize this. Can we reliably detect which babies are most vulnerable? What else should our readers know about your work? One is you can’t treat brain data agnostically. Two is that you have to account for significant heterogeneity, brain’s natural heterogeneity and heterogeneity in conditions. Brains are hard! That ’s my message. Brains are hard. [ laughs ] You worked at Imperial and now you are working at KCL. Can you tell us how it is to work at these institutions? What are the advantages of working in that kind of setting? King’s is a really exciting place to be right now. There’s a huge investment in healthcare technology research with some really great minds. If somebody wanted to develop training or healthcare technology, or machine learning for healthcare, it ’s absolutely the right place to be right now. They’ve got so much investment, huge GPU clusters, great people, diverse skills, a great marriage between clinical knowledge and technical knowledge, everyone’s working collaboratively. It ’s part of the golden triangle. We still work with Imperial. We work with Oxford. Everyone’s working together. It ’s a great place to be. 244 Women in Computer Vision