Detecting Cancer More Efficiently
Traditional methods of detecting cancer require a great deal of manual effort. Many of these techniques are inaccurate and depend on biomarkers that only become prevalent in later stages. Overdiagnosis is another ongoing issue, whereby false positives lead to further investigation and put additional anxiety and financial burden on patients.
To improve the state of cancer screening, the analysis of DNA methylation profiles on a whole-genome scale holds significant promise. By overcoming obstacles presented by large datasets deep learning methods give a major boost to this type of analysis.
This paper shows how a deep learning model, based on a convolutional neural network (CNN), can classify the cancer of a new DNA methylation profile using learning from publicly available DNA methylation datasets