Artificial intelligence-enabled liquid biopsy in cancer: a systematic review and meta- analysis of diagnostic performance and biological implications
Abstract
Background: Liquid biopsy offers a minimally invasive approach for cancer detection and monitoring, but its diagnostic performance is often limited by low signal abundance and biological noise. Artificial intelligence (AI) has been proposed as a computational framework to integrate heterogeneous and biologically noisy circulating biomarker data; however, the biological underpinnings, clinical relevance, and translational meaning of the reported performance gains remain unclear.
Methods: We conducted a systematic review and random-effects meta-analysis of studies published between 2022 and 2025 applying AI-based methods to liquid-biopsy data for oncologic diagnosis. Extracted outcomes included AUROC, sensitivity, specificity, cancer type, analyte class, and AI model. Pooled AUROC and absolute AUROC differences between AI-based and non-AI approaches were estimated with 95% confidence intervals using a random-effects model. AUROC values were logit-transformed prior to pooling to stabilize variance and were subsequently back-transformed for presentation. Statistical heterogeneity was quantified using the I² statistic.
Results: Twenty-eight studies across multiple tumor types met inclusion criteria, of which ten provided extractable data for quantitative synthesis. AI-enhanced liquid biopsy achieved a pooled AUROC of 0.924 (95% CI, 0.879-0.953). When head-to-head comparisons were available, AI-based models demonstrated an absolute AUROC improvement of 0.025 (95% CI, 0.019-0.030)…
How strong is the evidence?
Evidence Hub reads this as a systematic review. Its evidence level is 1 of 5 (Highest · pooled evidence). Pooled evidence from several studies sits at the top of the evidence pyramid.
Read the paper
Related studies
- Active Learning during Federated Learning for Object Detection
- Fairness-aware federated learning
- Decentralized Federated Learning
- Building a Personally Identifiable Information Recognizer in a Privacy Preserved Manner Using Automated Annotation and Federated Learning
- Preservation of Patient Level Privacy: Federated Classification and Calibration Models
- Efficient explainable federated learning
Go deeper in Evidence Hub
Evidence Hub adds a study data sheet (design, sample size, evidence level, key numbers), an AI summary and a chat grounded in the paper's text.