From models to tools: clinical translation of machine learning studies in psychosis
Andrea Mechelli, Sandra Vieira
Abstract
The past decade has seen a proliferation of neuroimaging-based machine learning studies in psychosis. 1 Furthermore, within the span of ten years, small local studies with few dozen participants have evolved into large multi-centre studies with several hundreds of participants. 2 , 3 , 4 , 5 In the midst of the search for accurate models, much attention has been given to methodological challenges including the impact of sample size, 6 , 7 the limitations of traditional case–control designs, 8 , 9 how to best deal with confounding variables 10 and the effects of heterogeneity 11 , 12 and inter-scanner variability, 13 just to mention a few. Although there are still important methodological challenges to overcome, substantial progress is being made, and a solution to these challenges is now considered to be a matter of when rather than if. 14 , 15 Wider discussions in the medical community about the ethical and legal implications of integrating machine learning models within diagnostic and prognostic assessment of patients are also underway. 16 , 17 , 18 , 19 , 20 Taken collectively, the progress being made towards the development and validation of neuroimaging-based machine learning models is encouraging, as if the different pieces of a very complex puzzle were slowly coming together. Less discussed however, are the challenges related to the development and validation of machine learning-based clinical tools. Here the critical distinction is between “models”, which tend to be developed and validated using a limited number of well characterised datasets with the aim of maximising accuracy, sensitivity and specificity, and “tools”, which must be feasible, acceptable and safe, and provide information that will guide clinical decision-making in real-world settings. This is a timely discussion, as a new generation of multi-centre studies aiming to develop machine learning tools to manage patients with psychosis is emerging (e.g., PSYSCAN, 21 PRONIA— www.pronia.eu ).