New Paper Biological methods and protocolsAccording to a paper published by Oxford University Press, doctors may soon be able to use artificial intelligence (AI) to detect, diagnose and treat cancer in their patients at an early stage. Cancer remains one of the most challenging human diseases, affecting more than 19 million people and killing 10 million every year. The evolutionary nature of cancer makes it difficult to treat advanced stage tumors.
Genetic information is encoded by the pattern of four bases (represented by A, T, G, and C) that make up the structure of DNA. Changes in the environment outside the cell can alter some of the DNA bases by adding methyl groups. This process is called “DNA methylation.” Each cell has millions of these DNA methylation marks. Researchers have observed changes in these marks in the early stages of cancer, which could help with early cancer diagnosis. They can look at which bases in DNA are methylated in cancer and how methylated they are compared to healthy tissue. Identifying the specific DNA methylation signatures that indicate different cancer types is like looking for a needle in a haystack. Researchers involved in the study believe AI can help.
Researchers from the University of Cambridge and Imperial College London combined machine learning and deep learning to train an AI model to look at DNA methylation patterns and identify 13 types of cancer (including breast, liver, lung and prostate cancer) from non-cancerous tissue with 98.2% accuracy. The model relies on tissue samples (not DNA fragments in blood) and will require additional training and testing on a more diverse collection of biopsy samples before it is ready for clinical use. The researchers here believe that a key aspect of this work is the use of an explainable and interpretable core AI model, which provides insight into the reasons behind predictions. The researchers explored the inner workings of the model and showed that it enhances and improves our understanding of the underlying processes that contribute to cancer.
Identifying these abnormal methylation patterns, possibly from a biopsy, could enable healthcare professionals to detect cancer earlier, which could dramatically improve patient outcomes, as most cancers can be treated or cured if detected early.
Computational methods like this model, through better training on more diverse data and rigorous testing in the clinic, could ultimately provide AI models that can help physicians detect and screen for cancer earlier, resulting in improved patient outcomes.”
Shamis Samarajwa, lead author of the paper
sauce:
Oxford University Press (USA)
Journal References:
NewshamI., others(2024) Early cancer detection and diagnosis through interpretable machine learning revealing cancer-specific DNA methylation patterns. Biological methods and protocols. doi.org/10.1093/biomethods/bpae028.
