Liquid biopsy has emerged as a powerful tool for cancer early diagnosis, prognosis, and treatment monitoring across a wide range of cancer types. The non-invasive collection of blood markers enables real-time insights into the disease biology. Cellfree circulating DNA (cfDNA) offers a potential window into various biological and genetic processes, especially circulating tumor DNA directly originated from tumor cells. Considering the attributes of cfDNA data, their inherent complexity, and the ease of collecting them over time, employing statistical modeling analyses appears necessary to extract relevant information. This review explores the diverse modeling approaches used to analyze cfDNA in oncology, emphasizing its role in oncology. After an overview of the current knowledge of cfDNA biology, the use of statistical analysis, machine learning, and non-linear mixed effects models is detailed for their application in clinical studies, both in processing cfDNA data at a particular time point and in capturing their temporal dynamics. Overall, this review provides a comprehensive overview of the diverse modeling approaches applied to cfDNA in oncology, with a focus on dynamic approaches.