Dr Jake Lever he/him

Our research focuses on applying machine learning and natural language processing to automatically gather and apply biomedical knowledge. Essentially we want to teach a computer to read all biomedical research literature out there and make useful suggestions for new research ideas. There’s been an incredible acceleration in artificial intelligence methods that can deal with text and we’re excited to apply them to real problems that could help scientists make sense of the overwhelming amount of research papers out there.

This research spans across methods for information extraction where we need to know which words are describing key entities such as drugs and diseases and what the text says about their relationships. From this we are able to build knowledge graphs that link these entities through knowledge unlocked from thousands of research papers. Knowledge graphs can be used to predict new associations and help explain existing knowledge. We’ve also been focussing on challenges specific to clinical text. Hospital discharge letters contain a wealth of information that could help scientists spot patterns of treatment effects and outcomes and elucidate underlying reasons behind diseases. We’ve been examining methods for generating realistic artificial records that can be used to build machine learning systems but don’t risk patient privacy. Potential CDT projects could span all of these areas from improving the underlying methods to applying them for a cancer or virology application with appropriate collaborators.

Our research relies heavily on collaborations as we want to develop methods that are invaluable for biomedical scientists. These have focussed on projects with biomedical research groups. We have existing collaborations with research groups in the US gathering knowledge about cancer. We’re also interested in how humans work with these systems and some of the big questions of explainability: how would a computer explain a research idea and convince scientists to invest time and money to test it.

Positive research culture is an integral part of a successful research group. Learning to be a good researcher is also learning to be honest, creative and supportive of others. Discovering new ideas can be one of the most exciting and fulfilling careers in the right environment with the right colleagues. Diversity of colleagues brings out the best discussions and ensures that everyone is able to contribute. EDI efforts are key to this. To help make sure that we have the right environment, I am a member of the school’s EDI committee and am part of initiatives to improve the culture of the school. I have also been part of the Glasgow Crucible programme to learn more about developing good research culture. Together we can ensure that everyone is able to pursue their research under the best conditions.