Opioid Use Disorder Prediction Using AI and Existing Risk Models

Explore a comparative effectiveness study on predicting opioid use disorder (OUD) using artificial intelligence and existing risk models in this 52-minute conference talk by Sajjad Fouladvand from Stanford University. Delve into the development of AI-based models for OUD prediction and their performance compared to clinical tools like the unweighted opioid risk tool. Examine the methodology, predictors, data formatting, and various models employed, including random forest, modified transformer, and logistic regression. Gain insights into the results, which demonstrate the superior predictive capabilities of AI algorithms, particularly the transformer-based model. Consider the potential impact of integrating AI into clinical care for improved risk stratification and patient management in opioid treatment. Conclude with a discussion on future projects and engage in a Q&A session to further explore this critical healthcare application of artificial intelligence.

Introduction
Background
Methodology
Predictors
Treatment
Data Formatting
Models
Random Forest
Modified Transformer
Opioid Risk Tool
Data
Results
Attention
Conclusion
Future Project
Questions