Abstract
<jats:p>Liquid biopsy technologies—spanning cell-free DNA (cfDNA), circulating tumor RNA (ctRNA), extracellular vesicles (EVs), and circulating tumor cells (CTCs)—offer a non-invasive window into somatic genome dynamics. However, translating high-throughput biofluid diagnostics into clinical oncology remains bottlenecked by fragmented analytical pipelines, ad-hoc bioinformatics, and siloed inference models that lack systemic interoperability, continuous feedback, and causal grounding. This framework paper (paper 1) proposes a canonical 7-layer reference architecture for AI-enabled liquid biopsy systems. Designed specifically to bridge molecular biology and AI systems engineering, this framework defines the data contracts, algorithmic paradigms, feedback loops, and governance mechanics required for end-to-end multi-omic reasoning. We present an indepth technical formulation of biological foundation models (Layer 3), complete architectural specifications with diagram layouts, comprehensive regulatory mapping tables (FDA PCCP, EU AI Act, IVDR), and operational blueprints for future autonomous "intelligent laboratories."</jats:p>