PI: Liz Washbrook

University Research Fellowship: August 2018 – July 2019

This project aims to improve our understanding of the risk factors that link mental health problems in early adolescence to poor outcomes such as failure to pass exams, become engaged in employment, education or training, or enroll in higher education. The research involves the application of a computationally-intensive technique – machine learning or ML – that has traditionally been used in computer science but has only very recently been applied in the social sciences. ML involves the use of algorithms that “learn” from a high-dimensional set of input data to build a data-driven predictive model. It is an ideal technique when the purpose of the research is prediction (here of young adult outcomes) on the basis of a large volume of data (here children’s mental health symptoms and circumstances in the first years of secondary education) when there are likely to be complex interactions between multiple predictive influences.

The research will draw on two large longitudinal studies: the ALSPAC cohort  of children born in the south-west of England in the early nineties (N=14,000) and the Add Health cohort of children in the US who were born in the late seventies and early eighties (N=20,000). Methodologically, these analyses will give new insights into the generalisability of ML models in the context of adolescent health and education across different education and public health systems. In terms of practice, a goal of the research is to improve the targeting of school-based mental health services through the development of tools that rely only on simple paper-based screeners for mental health symptoms and indicators that are readily available to teachers. This is a 12-month project supported a University of Bristol Institute for Advanced Studies University Research Fellowship (URF). It builds on a current ESRC-funded grant (2017-2019) Mental health and educational achievement in UK adolescents on which I am a co-investigator along with colleagues from the University of Bristol School of Population Health Sciences.