Toward trustworthy artificial intelligence in multi-omics: a review of reproducibility, stability, and interpretability

The integration of multi-omics data has become increasingly important in advancing precision medicine and systems biology. However, the reliability and trustworthiness of artificial intelligence (AI) models applied to such data remain critical concerns. This review examines the evolution and current landscape of reproducibility, stability, and interpretability in AI-driven multi-omics analysis. We explore these three pillars of trustworthiness in recent literature, with a particular focus on methodological innovations, benchmarking practices, and biological relevance. Drawing from key publications, including those featured in Briefings in Bioinformatics, we highlight emerging frameworks that aim to make multi-omics models more robust, transparent, and translationally meaningful. We advocate for routine adoption of TRUST-aligned evaluation practices, including structured stability assessments, multi-cohort benchmarking, and standardized model-card reporting, as default components of future multi-omics AI development. We conclude by outlining key challenges and future directions for developing trustworthy AI systems capable of supporting reproducible, interpretable, and clinically meaningful multi-omics research.