A. Calviño, A. Moreno-Ribera, S. Pineda
Big data has revolutionized the biomedical field and the way we study complex traits resulting in the generation of vast amounts of omics data, such as genomics or immunomics. In this context, studies using classical statistical techniques are becoming too simplistic when considering them.
Instead, Machine Learning techniques, that may reflect combinatorial effects and can deal with datasets with more variables than observations, should be contemplated to address such complexity. However, even when the most frequent supervised methods are applied, low predictive power can be achieved in some applications, such as cancer.
In this work we show how association rules can be applied with predictive purposes to overcome the previous drawbacks. More precisely, we apply them to a real data set made of 107 tumoral pancreatic samples and 117,486 germline SNPs, showing the good performance of the proposed method to predict the immunological infiltration.
Keywords: Genomics; High-throughput data; Machine Learning
Scheduled
Biostatistics
November 8, 2023 5:20 PM
HC4: Sacristía Room