Abstract:181 Background: The detection of methylation in circulating free DNA (cfDNA) has become a pivotal approach for the early screening and diagnosis of cancer. Traditional methods, face challenges including high detection limits, inconvenience, and elevated costs. Our prior work developed a rapid, cost-effective, and sensitive electrochemical methylation detection method for multi-cancer early detection (MCED), yet it couldn't identify the tissue of origin (TOO). We aimed to enhance this method by integrating traditional tumor biomarkers with artificial intelligence (AI) algorithms for improved cfDNA methylation detection and rapid TOO determination. Methods: A training cohort of 626 individuals was analyzed, including 173 colorectal cancer (CRC) patients, 49 patients with hematological tumors, 273 patients with other types of tumors, and 131 healthy controls. This analysis utilized electrochemical detection, carcinoembryonic antigen (CEA), and carbohydrate antigen 19-9 (CA19-9). We evaluated the diagnostic performance of these markers using ten types of AI algorithms. The optimal algorithm was selected to construct diagnostic models for CRC and hematological tumors. These models were then validated in an external cohort comprising 100 cancer patients. Results: In the training cohort, the logistic algorithm outperformed others, achieving AUC of 0.915 for CRC, 0.850 for hematological tumors, and 0.860 for other tumors. Based on these results, we opted to utilize the logistic algorithm for the development of diagnostic models for CRC and hematological tumors. The CRC model surpassed the performance of the electrochemical adsorption rate (AUC = 0.817) in CRC, though it showed limited efficacy for hematological tumors (AUC = 0.544) and other tumor types (AUC = 0.806). Conversely, the hematological tumors model achieved its highest performance in hematological tumors (AUC = 0.833), with commendable results for CRC (AUC = 0.831) and other cancers (AUC = 0.799) as well. In the validation cohort, the CRC model accurately identified 32 out of 35 CRC patients, and 17 of these were also detected by the hematological tumors model. Among the hematological tumors patients, 18 out of 20 positive with the hematological tumors model, and 15 were detected by the CRC model. For the remaining 45 patients with various other cancer, 22 tested positive with the CRC model and 28 with the hematological tumors model. Conclusions: Integrating machine learning with electrochemical cfDNA methylation detection and tumor markers provides a powerful approach for accurate cancer detection and TOO determination, promising to improve early diagnosis and personalized treatment strategies.

Incorporating increased variability in testing for cancer DNA methylation

Detecting Differentially Methylated Loci for Illumina Array Methylation Data Based on Human Ovarian Cancer Data.

Detection of differentially methylated CpG sites between tumor samples with uneven tumor purities.

Detecting differentially methylated regions in bisulfite sequencing data using quasi-binomial mixed models with smooth covariate effect estimates

Addressing dispersion in mis‐measured multivariate binomial outcomes: A novel statistical approach for detecting differentially methylated regions in bisulfite sequencing data

DNA Methylation Markers for Diagnosis and Prognosis of Common Cancers.

A novel artificial intelligence-enhanced electrochemical method for rapid methylation detection in cfDNA: Advancing multi-cancer early diagnosis.

A Hybrid Ensemble Approach for Identifying Robust Differentially Methylated Loci in Pan-Cancers

Abstract 3666: Multi-cancer early detection using metrics of DNA methylation based epigenetic instability

A new parsimonious method for classifying Cancer Tissue-of-Origin Based on DNA Methylation 450K data

Petm: a Penalized Exponential Tilt Model for Analysis of Correlated High-Dimensional DNA Methylation Data

Identifying Differential Methylation in Cancer Epigenetics via a Bayesian Functional Regression Model

Comparison of Different Normalization Assumptions for Analyses of Dna Methylation Data from the Cancer Genome

Integrative analysis identifies potential DNA methylation biomarkers for pan-cancer diagnosis and prognosis

Correlation-Adjusted Simultaneous Testing for Ultra High-dimensional Grouped Data

Advances in cancer DNA methylation analysis with methPLIER: use of non-negative matrix factorization and knowledge-based constraints to enhance biological interpretability

A Novel Tissue-Free Method to Estimate Tumor-Derived Cell-Free DNA Quantity Using Tumor Methylation Patterns

Reproducibility and Concordance of Differential DNA Methylation and Gene Expression in Cancer.

Predicting DNA methylation from genetic data lacking racial diversity using shared classified random effects

DNA Methylation Loci Identification for Pan-Cancer Early-Stage Diagnosis and Prognosis Using a New Distributed Parallel Partial Least Squares Method.

Application of the Rank-Based Method to DNA Methylation for Cancer Diagnosis.