Artificial intelligence and machine learning in cell-free-DNA-based diagnostics.

Area of Science:

• Genomics and Bioinformatics
• Computational Biology
• Molecular Diagnostics

Background:

• Circulating cell-free DNA (cfDNA) in plasma offers noninvasive diagnostic potential for fetal aneuploidies, cancer detection, and transplant monitoring.
• High-throughput sequencing enables detailed analysis of cfDNA characteristics, yielding numerous biomarkers across genetics, epigenetics, transcriptomics, and fragmentomics.
• Machine learning (ML) and artificial intelligence (AI) excel at integrating high-dimensional data, making them suitable for advancing liquid biopsy applications.

Purpose of the Study:

• To review and highlight various AI and ML approaches applied to cfDNA-based diagnostics.
• To discuss the integration of ML/AI with cfDNA analysis for noninvasive prenatal testing and cancer liquid biopsy.
• To explore future directions for leveraging cfDNA fragmentation patterns using ML/AI in methylomic and transcriptional investigations.

Main Methods:

• Introduction to the biology of cfDNA and fundamental concepts of ML and AI technologies.
• Discussion of selected ML/AI-based applications in cfDNA diagnostics, including noninvasive prenatal testing and cancer liquid biopsy.
• Analysis of specific applications such as fetal DNA fraction deduction, plasma DNA tissue mapping, and cancer detection/localization.

Main Results:

• AI and ML effectively analyze cfDNA characteristics, enhancing the precision of noninvasive diagnostic tests.
• Applications demonstrated include accurate fetal DNA fraction determination and improved cancer detection and localization through liquid biopsy.
• cfDNA fragmentomics, combined with ML/AI, shows promise for future diagnostic advancements in epigenetics and transcriptomics.

Conclusions:

• AI and ML are powerful tools for advancing cfDNA-based liquid biopsy, offering significant improvements in noninvasive diagnostics.
• The integration of ML/AI with cfDNA analysis holds great potential for early disease detection, risk assessment, and personalized medicine.
• Future research should focus on further developing ML/AI algorithms to fully exploit cfDNA fragmentation patterns for comprehensive diagnostic insights.