A Functional Taxonomy of Artificial Intelligence Methods in Liquid Biopsy

<jats:p>While early literature in AI-driven biofluid diagnostics primarily lists algorithms by machine learning topology (e.g., Random Forests vs. Convolutional Neural Networks), such taxonomic schemes fail to capture an algorithm's functional role within the clinical diagnostic workflow (AbdelHamid et al., 2026; Long &amp; Long, 2026; Sisca et al., 2026). Building upon our 7-layer reference architecture for AI-enabled liquid biopsy (Long &amp; Long, 2026), this paper (framework paper 2) presents an 8-level Functional Taxonomy of Artificial Intelligence Methods in Liquid Biopsy. Rather than categorizing models by their computational architecture, this framework classifies algorithms by their operational objective within the biofluid processing continuum: Level 1 (Data Acquisition AI), Level 2 (Quality-Control AI), Level 3 (Signal Enhancement AI), Level 4 (Feature Discovery AI), Level 5 (Prediction AI), Level 6 (Decision-Support AI), Level 7 (Generative AI), and Level 8 (Autonomous AI Agents). We demonstrate how this operational classification bridges wet-lab robotics, bioinformatic denoising, multi-omic latent space modeling, and closed-loop diagnostic orchestration (Long &amp; Long, 2026). Standardizing AI methods by functional utility provides a common vocabulary for bioinformaticians, machine learning engineers, regulatory bodies, and clinicians, facilitating the translation of multi-omic liquid biopsy systems into scalable precision oncology.</jats:p>