Using AI to detect and diagnose cancer in patients may soon allow for earlier treatment, say investigators at Cambridge University and Imperial College London.
The key is in our DNA
Genetic information is encoded in DNA by patterns of the four bases (A, T, G and C) that make up its structure. However, environmental changes outside the cell can cause some DNA bases to be modified by adding a methyl group (this process is called “DNA methylation”).
Each individual cell possesses millions of these DNA methylation marks. Researchers have observed changes to these marks in early cancer development and they could assist in early diagnosis of cancer. Currently, It’s difficult to examine which bases in DNA are methylated in cancers and to what extent, compared to healthy tissue.
So investigators trained an AI model, using a combination of machine and deep learning, to look at the DNA methylation patterns. It identified 13 different cancer types (including breast, liver, lung, and prostate cancers) from non-cancerous tissue with 98.2% accuracy. They found that the model reinforces and enhances understanding of the underlying processes contributing to cancer.
Additional training and testing
However, they say this model only relies on tissue samples (not DNA fragments in the blood), so it would need additional training and testing on a more diverse collection of biopsy samples to be ready for clinical use.
Identifying these unusual methylation patterns may allow healthcare providers to detect cancer early. “Computational methods such as this model, through better training on more varied data and rigorous testing in the clinic, will eventually help doctors with early detection and screening of cancers,” said the paper’s lead author, Shamith Samarajiwa, in a statement.
Citation: Newsham, I., Sendera, M., Jammula, S. G., & Samarajiwa, S. A. (2024). Early detection and diagnosis of cancer with interpretable machine learning to uncover cancer-specific DNA methylation patterns. Biology Methods and Protocols, 9(1). https://doi.org/10.1093/biomethods/bpae028