 Systematic reviews summarise all the best available research evidence to inform decision making and health. Research had been done that looked at whether steroids were helpful in improving the maturation of premature babies' lungs. The review showed conclusively that giving steroids saved lives among premature babies and that has probably saved many thousands of lives. For a typical review we might well have 20,000 citations to sift through as a team. It can take two, three years to complete. Project Transform is an artificial intelligence project that brings together the technology that's able to manipulate and analyse data very efficiently. We've been able to create a data set that can train the machine to help us do this task so that we can then better direct human effort where it's most needed. We're using Cortana Intelligence to enable us to develop and deploy these machine learning applications in the cloud. We've got a series of different classifiers where we narrow the scope of what a particular citation is looking at. What we're able to do is to be able to identify the research which is likely to be relevant for a particular review. I'm using Azure Machine Learning Studio to develop and deploy web services which make these classifiers available. We're then able to visualise the performance of a data set across the set of studies. We find most of the relevant citations much, much earlier in the process. What I love about all of this is that it's all about that partnership between the human and the machine. Artificial intelligence will allow us to answer more complex questions more efficiently and with more precision, improving decision making and therefore helping patient health.