From:  Colorectal cancer worldwide: epidemiological trends, economic burden, and the promise of AI-driven solutions

 CRC screening tests using blood-based markers with AI integration and cited performance.

Blood test name or vendorAnalytesTargetAI algorithm + input DataPerformance metricsReferences
mSEPT9ctDNACRC specificRandom forest (cfDNA methylation data)AUC 0.82–0.89; Sens 69–77%; Spec 88–92%[148, 149]
FreenomectDNA + proteinCRC specificEnsemble deep learning (multi-omics: cfDNA, protein)AUC 0.94; Sens 79–91%; Spec 91–96%[150, 151]
CancerSEEKctDNA + proteinMulti-cancerRandom forest + logistic regression (multi-analyte)AUC 0.94; Sens 69% (CRC); Spec 99%[152, 153]
GuardantctDNACRC specificTargeted NGS + ML classifier (proprietary)Sens ~91%; Spec ~94%[154, 155]
Grail (Galleri)ctDNAMulti-cancerDeep neural networks (cfDNA methylation)Sens 67% (CRC); Spec 99.5%; PPV 88.7%; early-stage sens ~27.5%[156165]
Clinical GenomicsctDNACRC specificLikely logistic regression/SVM on methylationSens 81%; Spec 91% (early trials)[166169]

CRC: colorectal cancer; AI: artificial intelligence; mSEPT9: methylated SEPT9; ctDNA: circulating tumor DNA; cfDNA: cell-free DNA; Sens: sensitivity; Spec: specificity; ML: machine learning.