# ArcDFI: Attention regularization guided by CYP450 interactions for predicting drug-food interactions

## TL;DR

ArcDFI is a new AI model that predicts how drugs interact with food by considering liver enzymes called CYP450, improving prediction accuracy and explainability.

## Contribution

ArcDFI is the first model to incorporate CYP450-drug interactions for predicting drug-food interactions, enhancing generalizability and model explainability.

## Key findings

- ArcDFI outperforms ten baseline models in predicting drug-food interactions under cold-drug and cold-food settings.
- The model's attention mechanism reveals molecular features linked to drug-CYP450 interactions and DFI predictions.
- Incorporating CYP450 data improves predictive generalizability and provides insights into drug-food interaction mechanisms.

## Abstract

CYP450 isoenzymes are known to be deeply involved in the formation of drug-food interactions (DFI). Previously introduced computational approaches for predicting DFIs do not take drug-CYP450 interactions (DCI) into account and have limited generalizability in handling compounds unseen during model training. We introduce ArcDFI, a model that utilizes attention regularization guided by CYP450 interactions to predict drug-food interactions. Experiments on DFI prediction—evaluated under stringent cold-drug and cold-food settings—show that our model outperforms ten baseline approaches, demonstrating the effectiveness of incorporating CYP450 interactions. Analysis of its attention mechanism provides insight into its current understanding of DCI and how they are related to its DFI predictions. To the best of our knowledge, ArcDFI is the first DFI prediction model that incorporates the concept of DCI, resulting in improved predictive generalizability and model explainability. ArcDFI is available at https://github.com/KU-MedAI/ArcDFI.

What we eat can significantly change how medications work in our bodies. Some foods can reduce a drug’s effectiveness, while others can cause unexpected side effects. These drug-food interactions are often overlooked, yet they are critically important for patient safety. One of the main ways these interactions occur is through liver enzymes known as CYP450, which are responsible for metabolizing most clinical drugs. When foods interfere with these enzymes, they can alter how drugs are absorbed, processed, and eliminated. Despite the importance of CYP450 in drug metabolism, no previous studies have incorporated CYP450 information into computational models for predicting drug-food interactions. In this study, we present a new artificial intelligence model called ArcDFI, which is the first to explicitly include CYP450-drug interaction data in its prediction process. Our model not only improves the ability to predict interactions involving new drugs or foods, but also explains which molecular features are most responsible for these effects. By uncovering the hidden impact of food on drug behavior, ArcDFI can support safer drug development and more personalized treatment decisions.