This research will investigate brain and digital measures as distinct subtypes and biotypes in anxiety and fear disorders, and whether clustered and identified subtypes are transdiagnostic rather than disorder specific. For example, we will use machine learning to identify amygdala pathology clusters on a sample of all fear and anxiety disorders, and then further evaluate how specific disorders distribute onto the amygdala clusters. Semi-supervised clustering in this way has found success in identifying unique biotypes in major depressive disorder (Tozzi 2024). Identifying specific biotypes of disease can allow for more precise identification of treatment modalities and targets, for example identified clusters of brain pathology could guide neuromodulation targets (Tassone 2024). There is likewise growing evidence that transdiagnostic evaluation of anxiety disorders may reveal shared pathology and an alternative to existing classification (Stade 2023).
Some specific research questions include:
– How do brain imaging pathologies and digital measures cluster across all anxiety and fear disorders using unsupervised and semi-supervised methods?
– How do specific disorders and specific symptoms map onto these identified clusters?
– Are there more novel ML models of imaging data that we can deploy to serve as better inputs into these clustering models?