My name is Dr Sarah Morgan. I work primarily at the Cambridge Department of Computer Science and Technology, as an Accelerate Science Research Fellow. I am also affiliated to the Cambridge Psychiatry Department, and to The Alan Turing Institute in London, so I can sometimes be found at one of those places!
My research applies machine learning, network science and Natural Language Processing to better understand and predict mental health conditions.
One of my main interests is using brain Magnetic Resonance Imaging (MRI) to study schizophrenia and other mental health conditions. In particular, MRI brain images can be used to investigate brain connectivity, by calculating MRI brain networks where nodes represent large scale brain regions and edges represent connectivity between brain regions. MRI brain networks from patients with schizophrenia often show altered connectivity patterns compared to healthy volunteers. My research explores both whether we can use these connectivity patterns to predict individual patients' disease trajectories, and what they can teach us about the biological mechanisms underlying schizophrenia.
I am also interested in using data science to investigate other aspects of mental health, for example using network science and natural language processing methods to study patients' speech.
There are broadly two types ways my research could be applied in future. The first is that developing a better understanding of the biological mechanisms underlying schizophrenia (for example from brain MRI) might help lead to new therapeutics. About 20-30% of patients with schizophrenia don’t respond well to current treatments, so new therapeutics could potentially be life changing for those individuals.
The other way in which my research could have real world applications is by identifying signals that can help predict or monitor disease outcome for patients with psychotic disorders. For example, we are currently exploring whether there are signals in speech data that can predict outcome for people who have some early stage symptoms of psychosis. If so, that could help clinicians target treatments better at patients who are likely to have poor disease outcomes.
I saw the Henslow Fellowship advertised by Lucy Cavendish College.
I had been working as a postdoc at the Cambridge Brain Mapping Unit in the Psychiatry Department for about a year when I applied for the Henslow Fellowship. I had lots of ideas about how the research I was doing could be extended (for example by applying a new method that had just been developed in Cambridge to construct structural brain networks for patients with schizophrenia) and I thought a Henslow Fellowship would give me time to do that. I also knew the research community at Lucy Cavendish College is friendly and supportive and I thought my work would fit in well there.
My Henslow fellowship was instrumental in giving me time to develop my expertise in brain imaging and network neuroscience. It also gave me the chance to meet researchers from other disciplines, both at Lucy Cavendish College and at the Philosophical Society. A lot of my research is highly interdisciplinary, pulling together ideas from Computer Science, Physics and Psychiatry, so having those interactions has been extremely helpful in forming new collaborations.
For my work with brain MRI, one of the biggest challenges is that magnetic resonance images only give us approximately millimetre resolution, whereas the biological mechanisms we’re interested in happen at a much smaller scale. We have recently started linking brain MRI data to genetic and genomic data to try to traverse these different scales. This sort of approach is quite new though, and we’re still learning the best ways to go about combining these different data modalities.
For my work with speech data, a key challenge is that there is relatively little data available at the moment. We are collaborating with clinicians who are hoping to start collecting more data soon, which is very exciting!
Show All
More powerful, longer-lasting, faster-charging batteries – made from increasingly more sustainable resources and manufacturing processes – are required for low-carbon transport and stable electricity supplies in a “net zero” world. Rechargeable batteries are the most efficient way of storing renewable electricity; they are required for electrifying transport as well as for storing electricity on both micro and larger electricity grids when intermittent renewables cannot meet electricity demands. The first rechargeable lithium-ion batteries were developed for, and were integral to, the portable electronics revolution. The development of the much bigger batteries needed for transport and grid storage comes, however, with a very different set of challenges, which include cost, safety and sustainability. New technologies are being investigated, such as those involving reactions between Li and oxygen/sulfur, using sodium and magnesium ions instead of lithium, or involving the flow of materials in an out of the electrochemical cell (in redox flow batteries). Importantly, fundamental science is key to producing non-incremental advances and to develop new strategies for energy storage and conversion.
This talk will start by describing existing battery technologies, what some of the current and more long-term challenges are, and touch on strategies to address some of the issues. I will then focus on my own work – together with my research group and collaborators – to develop new characterisation (NMR, MRI, and X-ray diffraction and optical) methods that allow batteries to be studied while they are operating (i.e., operando). These techniques allow transformations of the various cell components to be followed under realistic conditions without having to disassemble and take apart the cell. We can detect key side reactions involving the various battery materials, in order to determine the processes that are responsible ultimately for battery failure. We can watch ions diffusing in, and moving in and out of, the active “electrode” materials that store the (lithium) ions and the electrons, to understand how the batteries function. Finally, I will discuss the challenges in designing batteries that can be rapidly charged and discharged.
Musical instruments like the clarinet and saxophone do not obviously have anything in common with a bowed violin string. This talk will explore the physics behind how these instruments work, and it will reveal some unexpectedly strong parallels between them. This is all the more surprising because all of them rely on strongly nonlinear phenomena, and nonlinear systems are notoriously tricky: significant commonalities between disparate systems are rare. For all the instruments, computer simulations will be used to give some insight into questions a musician may ask: What variables must a player control, and how? Why are some instruments “easier to play” than others?
Please Note: Due to building works, the CPS office at 17 Mill Lane, Cambridge is now closed until further notice. Business operations as usual. Please contact us by email only: philosoc@group.cam.ac.uk
Cambridge Philosophical Society17 Mill LaneCambridgeCB2 1RXUnited Kingdom
Office Hours: (Temporarily closed)Monday and Thursday -10am-12pm and 2pm-4pm.
philosoc@group.cam.ac.uk