The opioid crisis in the United States is a significant public health challenge, leading to numerous deaths and costing the economy billions annually. Our project aims to develop and validate innovative models to predict the risk of opioid use disorder (OUD) in individuals considering or currently prescribed opioids. This effort focuses on three main areas:
1. Genomic Prediction Panel (MODUS): We will create a genetic screening tool using SNPs associated with opioid use and metabolism. This tool will employ machine learning to assess an individual’s risk of developing OUD based on their genetic profile. The model will be developed using the AllofUs dataset and validated using the VA Million Veterans Program and a prospective observational study.
2. Deep Learning Models: These models will integrate genomic data, clinical history, behavioral patterns, and social determinants of health to predict OUD. Utilizing advanced AI approaches, including foundation models, we will develop and validate these models using large national datasets like AllofUs and VAMVP, ensuring high accuracy and generalizability.
3. Microbiome Prediction Panel (MICROUD): This novel screening tool will predict OUD risk based on an individual’s gut microbiome signature. The model will be developed using data from the Human Microbiome Project and The Microsetta Initiative, and validated through a prospective observational study. Additionally, a biobank of samples will be created to facilitate further research into OUD-related biomarkers.
Aims:
* Develop and validate MODUS for genomic risk stratification of OUD.
* Create and validate deep learning models that integrate multi-omic data for precise OUD risk prediction.
* Develop and validate a microbiome-based prediction panel for OUD.
* Establish a multi-omic biobank for ongoing research into OUD biomarkers.
Rationale: Current methods for predicting OUD risk are inadequate, often failing to account for the complex interplay of genetic, environmental, and clinical factors. By leveraging advanced machine learning techniques and diverse datasets, our models aim to provide a more accurate and holistic assessment of OUD risk. This will enable precision medicine approaches in clinical settings, improving patient care and potentially reducing the incidence of OUD.
Project Duration: This project spans three years, with an optional fourth to complete data collection.
Public Health Impact: Our project seeks to transform how OUD risk is assessed and managed in clinical practice. By providing healthcare providers with tools to identify high-risk individuals early, we can improve decision-making around opioid prescriptions, enhance monitoring and intervention strategies, and ultimately reduce the burden of OUD on individuals and society.