Depression can present in many different ways, with people experiencing a wide range of symptom combinations (Buss et al., 2023; Fried et al., 2020; Lorenzo-Luaces, Buss, et al., 2021). Most research has studied this by counting symptom profiles, but this approach can oversimplify differences between individuals. We use methods from information theory to better measure how diverse depression symptoms are and to evaluate whether common subtypes, such as melancholic and atypical depression, actually make diagnoses more consistent. Across approaches, we find that these subtypes do not meaningfully reduce symptom heterogeneity. We also study whether different symptom patterns predict who benefits most from different treatments using machine learning. Overall, specific symptom combinations rarely predict treatment outcomes, with a few exceptions (e.g., positive emotionality, insomnia). Instead, greater overall symptom severity is a more consistent predictor of better outcomes with higher-intensity treatments. The projects include:
- Measuring depression heterogeneity using symptom profiles and information theory methods
- Testing whether DSM subtypes (e.g., melancholic, atypical) reduce symptom heterogeneity
- Using machine learning to examine whether symptom patterns predict treatment outcomes
- Comparing outcomes across treatments, including CBT, medication, positive psychotherapy, and interpersonal therapy
Identifying specific symptoms (e.g., positive emotionality, insomnia) that may influence outcomes - Examining how overall symptom severity predicts response to higher-intensity treatments