A Multivariate Hawkes Process for Detecting Individuals with Depressive Disorder

Machine Learning –> Temporal Point Process; Mental Health.

• Depressive disorder is one of the most prevalent mental illnesses among the global population. However, traditional screening methods require exacting in-person interviews and may fail to provide immediate interventions. In this work, we leverage ubiquitous personal longitudinal Google Search and YouTube engagement logs to detect individuals with depressive disorder.

• We collected Google Search and YouTube history data and clinical depression evaluation results from $212$ participants ($99$ of them suffered from moderate to severe depressions). We then propose a personalized framework for classifying individuals with and without depression symptoms based on a mutual-exciting point process that captures both the temporal and semantic aspects of online activities. Our best model achieved an average F1 score of $0.77 \pm 0.04$ and an AUROC of $0.81 \pm 0.02$.

• My Contribution: I was the first author of this paper. I simulated the interplay between the distributional semantics and the temporal stochasticity of online activities with a mutually exciting multidimensional Hawkes Process.
  
  Through clustering, I first identified a set of implicit topics and labeled all the online interactions. Then, by modeling the labeled longitudinal online interactions with the multidimensional Hawkes Process, I obtained promising results in detecting depression symptoms.
  
  

• Please see our lightning talk at the NeurIPS 2020 Workshop on Machine Learning for Public Health here. The paper can be found here.