Developing and Validating a Multi-Omic, Psychosocial, and Environmental Model of Individual Differences in Pain and Placebo Responses - This project investigates chronic pain conditions, particularly temporomandibular disorder (TMD), focusing on the variability in pain sensitivity, treatment responses, and comorbidities such as depression and anxiety. A key mechanism influencing pain experience is endogenous pain modulation (EPM), which involves neural pathways that inhibit nociceptive signaling. Placebo hypoalgesia, a form of EPM, is known to depend on psychological, genetic, and environmental factors. Although much is known about the factors contributing to placebo responses, the reasons for individual differences in placebo efficacy remain poorly understood. It is also unclear how EPM influences pain severity and chronic pain severity. Current knowledge includes the finding that pain catastrophizing dimensions (rumination, magnification, helplessness) mediate the relationship between psychological factors and chronic pain severity. Additionally, jaw function, psychological distress, chronic overlapping pain conditions (COPC), insomnia, and socioeconomic position (SEP) have been identified as influencing TMD severity and placebo responses. Despite this, much remains unknown about the interactions between omic variability, SEP, and placebo responses, and how these factors together shape individual pain outcomes. To address these gaps, the project uses machine learning (ML) approaches to uncover novel insights and develop personalized predictive models for pain management. The proposed ML strategies include: 1. Supervised ML (Random Forests, XGBoost): These models will be used to identify and predict pain phenotypes and placebo response variability based on clinical data (e.g., depression, anxiety, pain sensitivity) from 402 individuals with chronic TMD pain and 400 healthy controls. 2. Unsupervised ML (Gaussian Mixture Models, Hierarchical Clustering): This approach will explore how genetic factors, such as OPRM1 polymorphisms, interact with SEP (e.g., income, education) to influence pain severity and placebo responses. It aims to uncover hidden patterns of genetic and environmental interactions that affect pain experiences. 3. Graph-based ML and Deep Learning: These advanced models will be used to identify biological pathways related to placebo responses and pain sensitivity by analyzing transcriptomic and proteomic data, integrating molecular, immune-related, and genetic factors. By employing these innovative computational methods, the project seeks to move beyond population-level associations and develop personalized models that can predict placebo responsiveness and pain sensitivity. Ultimately, the goal is to improve pain management by identifying individuals most likely to benefit from placebo-driven analgesic mechanisms.
