Researcher perspectives: Sophie Fairweather

Sophie Fairweather is one of our PhD students. She is exploring whether we can predict the risk of someone developing anxiety and depression in childhood by using information we have about their early life. She is supervised by Professor Golam Khandaker and Dr Hannah Jones.

In July 2026 we spoke to Sophie about her academic journey, her PhD and what motivated her to try building a prediction model for mental health.

Moving into research from the pharmaceutical industry

Research had always been bubbling away in the background for me.

After graduating, I spent 10 years in the pharmaceutical industry, working in medical information, scientific advisory and, later, research and development.

In pharma, there was a lot to keep me interested, but I got to a point where I wanted to lead my own project, in an area I felt passionately about, and follow it through over a longer period. I wanted to do something more hands-on, where I could be involved in everything: the research, the writing, the project management and the bigger picture.

I also knew I wanted to do research that was translational. That was one of the reasons I was drawn to the NIHR Biomedical Research Centre: Bristol’s approach. I liked the focus on research that has the potential to move towards real-world benefit.

Research focus – risk prediction and early-life anxiety and depression

My PhD is about risk prediction and early-life anxiety and depression. My overall aim is to develop a prediction model for identifying children with early signs of anxiety and depression. These children may not yet have a diagnosis but may be at higher risk of developing persistent anxiety and depression, beginning in childhood and continuing into adolescence.

A prediction model takes lots of different pieces of information and uses them to estimate risk. For example, information about the child’s school and home environment. In this case, the question is: can we use early-life information to predict which children are most likely to experience persistent anxiety and depression over time?

I am particularly interested in that persistent group because, in the future, this kind of model could potentially help identify young people with the greatest need. If we can identify children earlier, there may be opportunities to offer support earlier. For example, we could offer children help at school or provide parents with training on how to support a child with mental health concerns.

I am developing the model using the Millennium Cohort Study, which is a UK-representative dataset. I then plan to validate it in Avon Longitudinal Study of Parents and Children (ALSPAC) and Born in Bradford, which are different to one another with respect to the ethnic groups and socioeconomic factors they represent. Validation is important because I want to see whether the model holds true across different populations.

Predicting trajectories of depression and anxiety over time

I began my PhD with a systematic review of models predicting depression and anxiety trajectories over time. This showed a clear gap: few models focused on childhood and adolescence, and most existing work centred on adult depression, often looking at relapse or remission in people with an existing diagnosis.

I have also been modelling early-life emotional symptom trajectories across the Millennium Cohort Study, ALSPAC and Born in Bradford. This part of my work identified a persistent high-symptom group among children aged 3 to 14 in all 3 cohorts, representing around 5% of the population.

This group was strongly associated with maternal mental health, and deprivation. Children in more deprived groups were more likely to have persistently high symptoms.

We also saw cohort differences: in the younger cohorts (born after 2000), high symptoms appeared from an early age and persisted, while in ALSPAC (born in the 1990s) symptoms started low and increased suggesting possible earlier onset in more recent generations.

In the younger cohorts, we also saw that low-to-medium-level symptoms increased over time and eventually overlapped with high-level symptom trajectories but remained low and stable in the older cohort (ALSPAC).

Why mental health research?

I have always been interested in mental health, neuroscience and psychology. My mum works with people who have neurodevelopmental and mental health conditions. She always taught me to try to understand people. That became a core value for me, and I think it is what led me first to neuroscience and now to mental health research. Growing up, I was interested in understanding why people are the way they are.

Impact – why this research matters for the benefit of children and young people

The long-term hope is that this research could help develop ways to identify children and young people who might need early support with their mental health.

If a model like this could eventually be used in a community setting, such as a school or GP practice, it might help professionals understand which young people are at higher risk of persistent problems and prioritise support earlier. Ultimately, I hope my work benefits children and young people. I also hope it would be useful for schools, parents, mental health teams and other professionals trying to identify young people in need of support.

That said, there are important ethical questions. If you flag a child as high risk very early on, what does that mean for them? How is that communicated to the child and their family? What happens if the model incorrectly identifies someone? Could labelling someone as high risk have unintended consequences, or even become a self-fulfilling prophecy?

In the future, I would like to do some community involvement and engagement work with teachers, parents and young people. Speaking with these groups would help me understand how realistic and acceptable this kind of model would be. It would also help me understand how it could be implemented to address some of the important ethical issues I’ve talked about.

Looking ahead

I am currently moving into the final stage of my PhD. I have identified a persistent symptom group and am now starting to build the prediction model. This means cleaning the data, looking at predictors and working through the modelling process.

After that, I hope to spend some time thinking about real-world application. More broadly, I hope to continue working in research that is translational and has a clear connection to real-world benefit.

I came into this area because I wanted to understand people and help people. That is still what motivates me now.