Generation of Drug-Induced Cardiac Reactions towards Virtual Clinical Trials

2025-02-11

AI Summary

Existing generative models struggle to faithfully model individualized drug effects on electrocardiograms (ECGs), contributing to high failure rates and exorbitant costs in cardiac drug clinical trials. To address this, we propose the Drug-Aware Diffusion Model (DADM), a novel framework designed for virtual clinical trials. DADM is the first to embed a physics-informed, physiology-based ECG ordinary differential equation (ODE) model into a diffusion architecture, enabling synergistic integration of physiological priors and data-driven learning via a dynamic cross-attention mechanism. Furthermore, we extend ControlNet to jointly encode demographic covariates and multidimensional drug features—such as dose, pharmacokinetics, and molecular properties—to support fine-grained, patient-specific ECG response simulation. Evaluated on two large-scale real-world ECG databases covering eight classes of cardiac drugs, DADM outperforms eight state-of-the-art methods, achieving ≥5.79% improvement in drug-response classification accuracy and an 8% gain in recall.

Technology Category

• Computer Vision: Diffusion Models for Vision
• Machine Learning: Deep Generative Models & Autoencoders
• Cognitive Modeling & Cognitive Systems: Simulating Human Behavior

Application Category

• User Modeling, Personalization and Recommendation: Accountability, Transparency, and Ethics for personalization
• Economics, Online Markets and Human Computation: Economic ramifications for generative AI infrastructure and applications
• Graph Algorithms and Modeling for the Web: Foundation models and LLMs for Web-related graphs

Abstract

Clinical trials are pivotal in cardiac drug development, yet they often fail due to inadequate efficacy and unexpected safety issues, leading to significant financial losses. Using in-silico trials to replace a part of physical clinical trials, e.g., leveraging advanced generative models to generate drug-influenced electrocardiograms (ECGs), seems an effective method to reduce financial risk and potential harm to trial participants. While existing generative models have demonstrated progress in ECG generation, they fall short in modeling drug reactions due to limited fidelity and inability to capture individualized drug response patterns. In this paper, we propose a Drug-Aware Diffusion Model (DADM), which could simulate individualized drug reactions while ensuring fidelity. To ensure fidelity, we construct a set of ordinary differential equations to provide external physical knowledge (EPK) of the realistic ECG morphology. The EPK is used to adaptively constrain the morphology of the generated ECGs through a dynamic cross-attention (DCA) mechanism. Furthermore, we propose an extension of ControlNet to incorporate demographic and drug data, simulating individual drug reactions. We compare DADM with the other eight state-of-the-art ECG generative models on two real-world databases covering 8 types of drug regimens. The results demonstrate that DADM can more accurately simulate drug-induced changes in ECGs, improving the accuracy by at least 5.79% and recall by 8%.

Problem

• Simulating individualized drug-induced cardiac reactions
• Improving fidelity in ECG generation models
• Reducing risks in clinical trials with in-silico methods

Innovation

• Drug-Aware Diffusion Model (DADM)
• Dynamic Cross-Attention (DCA) mechanism
• ControlNet extension for drug data