In the field of medicine, early detection of cancer has always been an urgent challenge. Now, a new study reveals the potential of artificial intelligence (AI) in this area. A research team from Imperial College London and the University of Cambridge has developed an AI model called EMethylNET, which can identify up to 13 different types of cancer, including breast cancer, liver cancer, lung cancer, and prostate cancer, with an accuracy rate of up to 98.2% by analyzing DNA methylation patterns in non-cancerous tissues. The findings of this study have been published in the journal "Biology Methods and Protocols" under the title "Early detection and diagnosis of cancer with interpretable machine learning to uncover cancer-specific DNA methylation patterns". The research demonstrates the enormous potential of AI in early cancer detection, although the model is still in the experimental stage. The researchers emphasize the importance of using interpretable AI models to understand their predictive logic, and their model has made significant progress in understanding the carcinogenesis process. The EMethylNET model performs well in multi-class classification, with an accuracy rate exceeding 98%. This achievement is of great significance for early cancer screening. The research team used machine learning methods to identify cancer-specific changes from normal tissue-specific methylation, involving complex data analysis and model training. They trained and evaluated various types of models, including logistic regression, support vector machines (SVM), gradient boosting decision trees (XGBoost), and deep neural networks (DNN), and ultimately determined that the XGBoost and DNN models performed best on independent datasets. The researchers also explored the internal workings of the AI model and found that multiple genes are closely associated with cancer-related processes. They conducted functional enrichment analysis on these genes and found that they play important roles in carcinogenesis and transcriptional regulation features, enriching in cancer-related pathways and networks. The corresponding author of the study, Shamith A Samarajiwa, stated that with better training on more diverse data and rigorous testing in clinical settings, such computational methods will eventually provide AI models that can assist doctors in early cancer detection and screening, leading to better treatment outcomes. In the future, this approach is expected to be extended to detect hundreds of cancer types and early detect multiple types of cancer through liquid biopsy methods. Additionally, this approach may also be used to screen for specific types of cancer or cancers of unknown origin, opening up new possibilities for early cancer diagnosis and treatment.