533 AI-driven predictive radiomics: A review of early detection methods for metabolic markers in preventative medicine

Abstract

Objectives/Goals: This review evaluates recent advancements in artificial intelligence (AI)-driven radiomics for detecting early metabolic and structural changes via imaging techniques like PET-CT and magnetic resonance imaging (MRI). It aims to assess AI’s potential to predict disease progression and explore its implications for personalized preventative medicine. Methods/Study Population: This review analyzed studies from the past five years that explored AI applications in radiomics for early disease detection. The studies primarily focused on patients at risk for metabolic, cardiovascular, and oncological diseases. AI algorithms, including deep learning models, were evaluated for their ability to detect subtle metabolic and structural changes in imaging data from modalities like PET-CT and MRI. We categorized methodologies based on imaging biomarkers targeted, AI model architecture, and the clinical populations involved. The review highlights the methods used across studies to assess AI’s effectiveness in predicting disease progression. Results/Anticipated Results: The review found that AI models consistently demonstrated superior performance in detecting early metabolic and structural changes compared to traditional radiology methods. Across multiple studies, AI was able to identify biomarkers associated with disease progression months before clinical symptoms appeared, particularly in metabolic, cardiovascular, and cancer patients. Deep learning algorithms showed high accuracy in analyzing imaging data, improving predictive outcomes. The findings suggest that integrating AI into clinical practice could enable earlier interventions, offering personalized preventative care, and reducing the progression of late-stage diseases. Discussion/Significance of Impact: AI-driven radiomics holds great promise for transforming healthcare by enabling life-saving early detection and precision-based interventions. This technology could significantly reduce mortality in diseases like cancer, heart disease, and diabetes by allowing for earlier, targeted preventative strategies.